Can I achieve a resolution down to +/-0.05deg with ADXL202?

Document created by analog-archivist on Feb 23, 2016
Version 1Show Document

Q

Can I achieve a resolution down to +/-0.05deg with ADXL202?

A

Theoretically this is possible. In practice, however, achieving a resolution
down to +/-0.05deg will be challenging. You are really asking about the
sensitivity of the accelerometer, that is, the smallest change in angle that
can be reliably measured. The sensitivity is limieted by the noise in the
accelerometer.

The noise in the ADXL202 is specified as 500ug/sqrt(Hz) typically. The total
noise therefore depends on the bandwidth. You as the system designer can
control the bandwidth of the accelerometer using the capacitors Cx and Cy.
Large values of Cx and Cy will reduce the bandwidth and reduce the noise. The
bandwidth is given by:
Noise bandwidth = 1.57 / (2 x PI x C x R)

Where C is the value of Cx or Cy, and R is the value of Rset which is nominally
32kOhm.

For a noise bandwidth of 1Hz, the total noise is 500ug, for a noise bandwidth
of 100Hz, 5mg.

You are measuring the angle of tilt by measuring the component of the
acceleration due to gravity which is along the measurement axis of the
accelerometer. The angle (theta)  and the measured acceleration (X) are related
by simple geometry as:
X = Sine (theta)

Assuming you can limit the noise bandwidth to 1Hz, we now convert the noise to
it's equivalent angle:
Theta = sin-1 (X) = 0.029 degrees.

So you can get the resolution you require. However, you'll need to calibrate
the ADXL202 ( as described in the datasheet and app notes), which normally
consists of exposing the accelerometer to +/-1G, recording the maximum
excursion and calculating the calibration co-efficients. You'll also need to
take care with the temperature drift which is typically 2mg/degC. You can
either recalibrate when ever there is a change in ambient temperature or store
different calibration co-efficients for a range of temperatures and measure the
ambient temperature.