Post Go back to editing

How to treat out-of-spec acceleration measurements

I am using a ADXL325 accelerometer to measure vibration of a moving vehicle, and have been getting acceleration values above the nominal ±6g range.  They are sporadic and are likely not critical to the overall measurement, but I am wondering what I should assume about the validity of these values and if I need to get a higher-rated sensor altogether.

Attached is an example plot showing this effect.



  • Just got off the phone with technical support, which confirmed that something is indeed wrong with the output ranges.  To clarify, the min/max output voltages I am reading from the sensor are 0.0838V and 3.2839V respectively, which using the ratiometric sensitivity value for Vdd=3.3V, 0.192V/g, would equate to min/max acceleration values of -8.1573g and 8.5098g respectively.  It was suggested that I put a buffer amplifier in between the output of the sensor and the analog input of the microcontroller I am using, which would make sense but I have not found any literature or other designs requiring this.  Does anyone have experience with this sensor, or could point me in the right direction?



  • Hi jonm,

    Your calculation above is correct.  The +/-6g is a nominal spec and based on the nominal sensitivity and the output range, you can record acceleration values as high as 8.5g.  However, a few things to consider:

    1. it is best to stay within 5% and ~93% of the output swing of the amplifier to prevent any non-linearities in the output data.  So for 3.3V operation, ideally should use 0.16V as the lower limit and 3.07V as the upper limit.

    2. the min/max acceleration values will also depend on the initial offset, which may deviate from ideal value 1.65V and initial sensitivity (192mV/g) can also differ from part to part.  The limits are specified in the datasheet.

    Typically, if sensor experiences out of range input acceleration you will notice that the output is clipped or truncated.