What parameters should be considered when choosing an accelerometer

Selecting the right accelerometer for your application can be tricky. To help you in selecting the right device for your application, here are some considerations for the technical aspects of these devices.

  1. Bandwidth (Hz): The bandwidth of a sensor indicates the range of vibration frequencies at which the accelerometer responds or obtains reliable readings. Normally, humans are unable to make body motions beyond the range of 10Hz to 12Hz. Therefore, a bandwidth of 40Hz to 60Hz is sufficient to sense tilt or human movement.

Please note that
In digital-output accelerometers, bandwidth is the highest frequency signal that can be sampled without aliasing by the specified Output Data Rate. Per the Nyquist sampling criterion, bandwidth is half the Output Data Rate.

In analog-output accelerometers, bandwidth is defined as the signal frequency at which the response falls to -3dB of the response to DC (or low-frequency) acceleration.

For example, ADXL314 is a digital-output accelerometer. It supports a wide bandwidth response and low power mode in measurements.

image
The bandwidth and output data rate of ADXL314

  1. Sensitivity (mV/g or LSB/g): Sensitivity is a measurement of the change in the output electrical signal for each of the smallest detectable change in signal or mechanical input. This specification only works at the single given test frequency.

For example, BMA530 supports different acceleration ranges, which leads to different sensitivities as shown in the below table.

image
Sensitivity of BMA530 under different acceleration rages.

  1. Noise Spectral Density (µg/√Hz): Noise Spectral Density is the power spectral density of noise or the noise power per unit of bandwidth. This noise varies corresponding to the reciprocal square root of the bandwidth. The faster we read accelerometer changes, the worse the accuracy. When the output signal is smaller under lower g conditions, noise has a greater impact on the accelerometer’s performance.

For example, ADXL357 has an ultra-low noise spectral density (all axes), 75 µg/√Hz.


Noise spectral density of ADXL357

  1. Zero-g Bias and Zero-g offset: Zero-g bias level specifies the sensor output voltage when there is no acceleration acting on the device. The actual zero-g level of an accelerometer may be different from the nominal value. Usually, the deviation from the ideal zero-g level is specified as a zero-g offset in the datasheet.

For example, the typical zero-g offset of BMA456 is around +/-20mg. This means that zero-g output typically varies in the range of +/-20mg around the expected ideal value.

image
Output Signal of BMA456

  1. Frequency response (Hz): This is the frequency range specified with a tolerance band (+/-5%, etc.) within which the sensor will detect motion and report the true output. Specified band tolerances allow the user to calculate how much a device’s sensitivity deviates from a reference sensitivity at any frequency within its specified frequency range.

  2. Dynamic range (g): This is the range between the smallest detectable amplitude that the accelerometer can measure to the largest amplitude before the output signal is distorted or clipped.