A follow-up question from the main log_308 analysis: now that EKFQ's β state is observable, would it make a better ball-driving signal than the lateral-G OnSpeed ships today? Same flight data; new strata.
The reverse-engineering audits at ~/code/onspeed/{g3x,dynon}-audit cover this:
- Garmin G3X: the slip-skid ball widget on the GDU is derived from body-frame lateral acceleration, not from a sideslip channel. Garmin's GSU 25C EKF doesn't appear to carry β as a state at all; β is reconstructed post-hoc when needed. The ball is only smoothed by the EKF Kalman gain on the body-Y accel state — no extra display-side LP. (
G3X_SMOOTHING.md:230-280) - Dynon SkyView: has a dedicated
SlipSkidBall::set_ahrs_dataclass on the GDU that reads theadahrs_data_tfield at offset +0x84. The audit's wire-format analysis shows it's **lat