Borgelt Dynamis variometer

On Thursday, March 21, 2019 at 8:58:32 PM UTC, Nick wrote:
Thanks.

There are standard algorithms such as Kalman filters.

https://en.wikipedia.org/wiki/Sensor_fusion

For example you could dead reckon from ASI, accelerators and gyros. It would drift over time, because its an integration. Combine that however with GPS, altimeter, etc, and you can end up with a system that has lower errors..

Gusts in one way, are just noise, and approaches like Kalman filters work well at eliminating noise. So if you fly into a horizontal gust, you get a change in ASI, but no appreciable longitudinal acceleration.

Nick, that approach sounds plausible and is very different from the simple pneumatic or internal electronic signal damping "gust filters" we use now. I can only imagine that the sensor fusion hardware you give a link to would need to be a bit more sophisticated than the MEMS chips in current vario/nav systems.

On the mems devices if you have played with them, are pretty standard.

Compasses have errors. The 3 axes are not always at 90 degrees. The center is not necessarily at 50%, and each axis may be scaled differently. It's quite straight forward to turn this ellipsoid into a unit sphere. That's just the calibration.

Same applies for accelerators. You can rotate and hold still to get the calibration.

Gyros are odd. They suffer the same but they drift over time, so you need to adjust for the drift. But over a short period they are good.

GPS is accurate certainly over long periods.

From what I know, TE and ASI, along with altimeters are noisy measuring devices. Noise changes pressure, and hence you get a random change in altimeter. Not sure how they are affected by temperature.

So what do you want to measure?

1. The orientation of the glider - bank, pitch and direction
2. It's velocity
3. It's acceleration
4. The vector of the air. Wind speed + vertical
5. Conceivably, slip.

6. Flaps
7. Conceivably even the control movements.

That should give you a good model, and you should be able to fit from that.

It is what the AHRS systems use anyway, for a subset. See the Madwick system

http://x-io.co.uk/open-source-imu-and-ahrs-algorithms/