When an Accelerometer Fails Calibration

A Practical Approach

Q:  I use this sensor on a daily basis, and haven’t noticed a difference in any of my measurement results.  Could it be that signals I am measuring do not have significant activity in the "failure" frequency range or that my measurements are not sensitive enough for me to have picked up on differences between the two sensors?

A:  Yes, it is quite possible.  It seems you have the benefit of some quantitative information in the calibration results, i.e., "X%."  If the usual behavior of the sensor was 4.9% at "Y" Hz and is now 5.1% (crossing a 5% criteria for an acceptable calibration), you would have a 0.2% change in your sensitivity, and your data.  This may or may not be detectable, depending on the predictability of your measured structure's behavior.

Q: How often do people usually get their equipment lab calibrated? 

A: It varies widely depending on the criticality of the measurement, the environments to which the sensor has been exposed, and the robustness of the sensor design.

An example of diverse criticality is, on one hand, an easily repeatable test on a test article that will be readily available later, versus a once-in-a-lifetime opportunity to test a structure, the behavior of which will have life and death impact.  Think automotive crash testing or destructive stress tests of prototype aircraft.

Examples of various environments will include a sensor that is used for a test and then stored, undisturbed, in laboratory storage in controlled conditions, versus a sensor that is mistakenly dropped from a 30-foot-high catwalk to a concrete surface.

Q:  If the accelerometer fails in a particular range of the spectrum, how likely is that failure to propagate to a wider portion of the spectrum?

A:  Again, it's a definite maybe.  It depends on the sensor design and the calibration data.  Please feel free to share the calibration data with us at The Modal Shop, Inc., and we’ll help you evaluate it.

All that being said, the calibration interval is frequently determined by a user company's internal quality system.  In some cases, the sensor manufacturer may even specify the calibration interval.

In all cases, please consider the importance of having the calibration performed according to well-established international standards, (16063-21 and -22 from ISO are the most widely accepted).  These standards provide multiple advantages, including global consistency of the acceleration measurements.  Neither of them specify pass/fail criteria, as another post suggests.  It is likely that the pass/fail criteria for your accelerometer matches the manufacturer specification sheet.

Also consider the importance of measurement uncertainty (or in other words, the accuracy) of the calibration service, and the accreditation of the service provider to ISO 17025.