How to Find an Accelerometer's Sensitivity

Attach the accelerometer to a data-recording device. The device could be a computer that automatically and continuously takes voltage readings, or it could be a voltmeter you can read and a pad of paper and pencil so you can write down the voltages.

Set the accelerometer with the axis to be measured oriented vertically. The acceleration on that axis is then 9.8 meters per second^2 or 1 g.

Take voltage readings from the accelerometer. You want to gather enough to get a good sample of the accelerometer's stability, but you've got to balance that against practical concerns. For example, if you're recording by hand, 10 individual readings is enough, while an automated system could let you record data every fraction of a second for five minutes.

Turn the accelerometer so the axis you're measuring is parallel to the force of gravity. The acceleration on that axis is now zero meters per second^2, which is 0 g.

Repeat the reading procedure from Step 3.

Calculate the average and standard deviation of the measurements taken at each accelerometer orientation.

Calculate the calibration curve for the accelerometer.

For example, if you measured an average of 840 millivolts (mV) in Step 3 and 90 mV in Step 5, then the slope of the calibration curve would be 1 g/(840 - 90 mV) = 1 g/ 750 mV = 0.00133 g/mV. The full equation is: acceleration = 0.00133 g/mV x (output - 90 mV).

Calculate the electronic noise of the accelerometer. The noise can be estimated by the standard deviation reading.

For example, the standard deviation of the 840 mV reading could be 7 mV, and the standard deviation around 90 mV could be 6 mV. Select the largest value as the conservative estimate of standard deviation.

Convert the standard deviation to a sensitivity. Use the calibration curve to make the conversion.

For example, the 7 mV standard deviation corresponds to an acceleration of 0.00133 g/mV x 7 mV = .0093 g. That's the sensitivity of your accelerometer.