Random Vibration Testing Standards

Recognize, please, that specifying random vibration tests only in terms of RMS g is not sufficient. A wide variety of spectra could all enclose the same area and thus provide the same RMS g. We must specify spectral details. That’s why we first focus on the frequency domain. Don’t expect to find simple spectra per the MIL standards on board real vehicles in the “real world”. “Real world” spectra, greatly affected by structural and panel resonances, tend to be very jagged. 

I hope you’re comfortable with spectrum tolerances given in dB units. If not, you might want to review ebook 1, section 1.27. We can never achieve exactly the PSD that the MIL standards show as straight lines. Our equalization is never perfect. How high above the 0 dB (specified) line may peaks go? How far below the line may notches go? The tolerance is usually stated in deciBels and a common value is ±3 dB. +3dB is a very wide tolerance (-50% +100%). It was appropriate for the old analog controllers. +1.5dB is more appropriate and more commonly seen currently.


22.1 Why are standards not effective? 

22.2 Military Standard 810, IEC, ASTM 

22.3 Earthquake (seismic) testing - NEBS 

22.4 How do we interpret what standards say? 

22.5 Focus on frequency domain 

22.6 “Real world” spectra not so simple 

22.7 Interpreting dB tolerances 

22.8 Shaker force requirement; spectrum area 

22.9 Multiple sine + random testing 

22.10 Analyzing narrow bandwidth peaks 

22.11 Spectral graphs misleading 

22.12 Statistical degrees of freedom 

22.13 Don’t forget the time domain 

22.14 Clipping 

22.15 Multiple zones