Using Random Vibration Data 2

[BRENTB]

At our company, we subject our electronic equipment to a specified random vibration PSD and monitor responses. Can you recommend a reference that will give practical (not a bunch of differential equations) ways to interpret this data to determine Equivalent Static Loads? If I could determine an equivalent Static Load, I could analyze the problem. Should I use the levels at the first mode or the grms under the entire PSD curve?

 

[CWM]

Look for a book by an author named "Dave Steinberg" with a book entitled something like "Vibration of electronic equipment". Not sure I am right on the title, but I know for sure the author is correct. This book should lead you in the right direction without the fog of theory typical with many other books. Steinberg is about as practical as you get.

I believe the method is to integrate under the resonant peak (PSD response curve) of interest using the half-power bandwidth (The interval containing the resonant frequency that starts at the frequency that has a PSD value half of the resonant peak [left side of resonant peak] and ends with the frequency that has a corresponding PSD value of again half the resonants peak's PSD value [right side of resonant peak]. Once you have integrated, most likely using numerical methods, you can take the square root of this result and you end up with a Grms value.

By knowing the Grms value and a resonant frequency value, you can calculate RMS deflection (See Steinberg's book for this relation). Then you can approximate peak deflection by multiplying the RMS deflection value by 3. I would strongly suggest you read throught the book first to verify the methodology.


Hope this helps.

CWM

 

[zahmad]

CWM is right in pointing you in the right direction i.e. Dave Steinberg and the title is "Vibratin analysis for Electronic Equipment". It is a John Wiley publications. The book has a wealth of knowledge.

As a first approximation, you can take the output Grms, usually printed on the plot, as a measure of the G loading. This is especially true if your equipment behaves like a single degree of freedom, with a strong resonant mode i.e. the first major mode. If this is true then most of the Grms i.e. the energy will come from the area under the resonant mode. Therefore integrating using the numerical methods using half power bandwidth is not necessary.

You will find useful information on how to calculate peak response, how to use in strength calculations, PCB deflections, component fatigue life and a lot more by reading through the above text book.

 

[goodvibes]

Brent,

RMS acceleration in g's is equal to the square root of the product of the total banwidth (in Hz) multiplied by the PSD level (in g2/Hz). Once you have the g rms value you multiply the static loads by this value to obtain the load experienced by the euipment at that PSD level. If your testing at a constant PSD over a range of frequencies (bandwidth) then the above calculation is valid.

If you are testing at different PSD levels at different bandwidth intervals you must do the above calculation for each area of constant PSD and add them to get the total g rms level. You are basically calculating the area under the curve for each section.

Sometimes specifications will say to test at a varying PSD level over a certain bandwitdth, and will call out a slope in DB per octave. Your equipment specification should have a graph showing the PSD vs freq that the equipment should be tested to.

the best "white paper" i have ever seen on this is subject is Wayne Tustin's paper on "understanding random vibration specifications". I have a very old copy (1963), but I think that he is still around and works for ENDEVCO. They might be able to send a copy of his technical paper.

I hope this helps! :