Posted on April 22, 2015 at 09:20We use an L3G4200D rate-gyro in our project.
Part of the setup involves acquiring calibration data from the sensor (zero-rate drift versus temperature as reported by the sensor itself).
After applying the collected calibration data in the application software we observe the following issue:
- Whenever we power-up and initialize the sensor there is a random chance to get a different zero-rate drift behavior compared to the calibration data previously collected
- There seem to be two different states, one which matches our calibration and another which does not
- The residual drift in the ''alternate'' state seems to be repeatable
We suspect the sensor may be randomly loading (or not loading) a factory calibration table on power up.
If we detect the ''bad state'' we can power-cycle the sensor one or more times (via CTRL_REG1) until we get the state which matches our calibration data.
However as you can imagine this is not really acceptable as the equipment needs to be static for the application software to measure the residual drift on startup.
The behavior has been observed on several sensors with varying magnitude.
The sequence we use for initializing the L3G4200D is as follows:
- Set data rate, power-up bit and X/Y/Z enable by writing 00111111 on CTRL_REG1
- Set full-scale angular rate by writing 10010000 to CTRL_REG4
- Enable FIFO by writing 01000000 to CTRL_REG5
- Set FIFO mode and watermark value by writing 01001010 to FIFO_CTRL_REG
Is this a known issue, and is there a way to predictably start the sensor always in the same ''calibration state''?
Thanks and kind regards
#l3g4200d-drift
