Jim Carr wrote:
> If, as you write below, the g-analyst does not do the correction
> for you, I would say that this may be why they have three
> accelerometers but that they may not make full use of them.
I can't speak for the internal use of the acceleromteres - Valentine's
documentation doesn't go into great detail - but it is true that the
g-analyst's output only includes lateral and longitudinal acceleration.
This is not just the sample stream from the "lateral" and "longitudinal"
accelerometers, however.
> Or perhaps I don't quite understand the calibration setup
> and what it does.
Any of the systems we're discussing have to either specify or identify
via calibration the relationships between the accelerometer's internal
coordinate system and the vehicle's body. The only thing you're required
to align accurately when you install the g-analyst accelerometer module
is its longitudinal axis with the car's. The unit may be rotated along
its longitudinal and/or lateral axes, and the calibration procedure will
determine the necessary rotation of axes to acquire an accurate
body-centered axis set for the car. You calibrate once with the car
parked, then repeat with the car in the same spot but pointing in the
opposite direction.
If this procedure isn't available, then you have to accomplish the
required alignment physically, which may not be very easy.
There are two sets of alignment issues: 1) accelerometer axes with
vehicle axes, and 2) vehicle axes relative to level. The g-analyst's
calibration procedure takes care of the first one, and three
accelerometers are required to accomplish this. I don't believe you can
perfectly account internally for pitch, roll, or yaw with only linear
accelerometers.
Jay
|