In the 3rd generation MTi-G, there was a setting "GPS lever arm". The goal of a GPS lever arm is two fold: re-targeting the position of the antenna to the position of the tracker and compensating for (centripetal) accelerations.
The compensation for acceleration has a very limited effect on the MTi-G performance. When an object is making a rotation with the antenna and MTi-G apart from each other, there is a discrepancy. However, going through the possible combinations, there are hardly any situations where this is an issue:
- When the yaw is changing fast (e.g. 50 º/s), the speed is probably low. For a significant centripetal force of 2g lasting ~2 seconds, the speed must be 150 km/u at a radius of 100 meter. Also, in this case, the gyros will coast for the 2 seconds.
- When the yaw is changing fast and the speed is close to 0 (e.g. hovering drones), there is no centripetal acceleration to compare.
Still, it is advised to keep the antenna of the MTi-G close to the MTi-G, especially in the horizontal direction.
The position of the antenna is the position outputted by the MTi-G. Position correction of the antenna to the tracker is indeed possible with a lever arm. However, the errors in timing, the navigation algorithm and GPS estimation itself are far larger than the recommended (or possible) lever arm. Consider the following errors:
- Timing: The MTi-G receives the GPS signals at a 4Hz frequency. As we don’t control when we start triggering the GPS receiver, there is a 250 ms resolution on the required time on when you want the position. At 30 m/s, this already results in 7.5 m uncertainty.
- Navigation algorithm: in order to get a smooth result, the navigation algorithm typically averages 2-3 epochs (GPS samples). This results in an extra GPS latency of ~ 0.5 seconds. The error in position at 30 m/s will be several meters: it is uncertain how the navigation algorithm ‘predicts’ where the next position will be. The MTi-G uses the inertial sensors to output position values with a far lower latency.
- Measurement noise on the GPS position estimation: even with SBAS, the uncertainty in measured position is 2.0 m CEP. This alone is already greater than typical lever arms.