Model aircraft helicopters have had gyros for decades already. Without a yaw gyro it is as good as impossible to fly one.
GPS based yaw measurements have been used extensively in automotive fields for a long time – chassis dynamics, motorsport, etc. They use two separate GPS antennas at opposite ends of the vehicle.
But this story still doesn’t add up. A yaw damper only needs a yaw rate input, not a full attitude source. And the Phenom will have dual (or triple?) AHRS, which will each work without one of either GPS or air data anyway.
If for some reason the yaw damper can’t work with a AHRS that is degraded due to loss of GPS data (slower yaw data output rate?) then it should just be using the raw information from the rate gyros instead.
Peter wrote:
I guess GPS is a lot cheaper than a yaw gyro…
Not really, given that both are contained in fairly low end phones already…
But GPS gives you absolute measurements, while the yaw gyro gives you only the rate with a fairly high and variable bias…
Airborne_Again wrote:
How is that possible? A GPS gives you position, not attitude.
You need multiple (2) antennas. You could compute the positions of the two antennae, and if you know where they’re mounted on your rigid body, that gives you the attitude of the body. If you do it a bit more cleverly than just computing two coordinates, most of the error sources cancel out, so it can be fairly accurate.
And given that the wavelength of an L1 GPS signal is about 20cm, they don’t even need to be far apart, if your receiver can do carrier phase correlations.
In a queer and cruel (read british) way it would be fun to do the same with a bunch of NDB signals – if it wasn’t for their impractically large wavelength (~600m), and AFAIK there are only very weak guarantees about their carrier wave stability. (And the fact that there are almost no NDB left outside the UK)
tomjnx wrote:
But GPS gives you absolute measurements, while the yaw gyro gives you only the rate with a fairly high and variable bias…
Care to explain why bias is of any practical relevance when you are only interested in the rate (to stabilize yaw), with a MEMS rate gyro?
tomjnx wrote:
But GPS gives you absolute measurements, while the yaw gyro gives you only the rate with a fairly high and variable bias…
Which are solved problems. Model helicopters were mentioned already. The gyro for a model heli in its most basic form is just to control the tail. Modern ones are a MEMS gyro and they can keep a rock solid hold on the tail even if you bang the collective from stop to stop without touching the tail rotor channel. It does seem a bit odd that the yaw damper can’t be run as an independent system and would require a GPS input when it’s what MEMS gyros seem to be best at.
An AHRS has to keep track of which way is up, and which way is North. MEMS systems have terrible long term drift of rotation rates (it varies hugely with temperature) so the local acceleration vector is fed into the computations as a long term average.
This is no different from your gyroscopic AI having a special mechanism slowly to level itself to correct for precession and to self erect after an attitude upset. It’s typically about 5 degrees per minute that it moves back to the “vertical”.
During a turn, of course the local acceleration vector isn’t vertical any more, and your AI integrates an increasing N/S and/or E/W error into the spin axis during a turn. The precession rate is low so the erection rate can be small, and the total error never gets to more than a few degrees. You can’t, after all, spend a very long time accelerating in one direction in an aircraft.
An AHRS may have to use higher “reverticalisation” rates to overcome higher drift rates in the MEMS than the slow precession of the mechanical gyro. In this case we can use GPS to measure the ground track of the aircraft and account for how the locally sensed acceleration differs from the vertical, so that the computed horizon level doesn’t “lean into” a banked turn, and you don’t get a pitch error when slowing or speeding up.
Either way, if you’re building a computer based attitude solver you feed all the information you have – including GPS – into it, and use a model of the aircraft dynamics to give you the most accurate attitude solution taking into account everything you know and the noise and accuracy of each information source. (This is what a Kalman filter does, at a basic level). By definiton, if you remove one source of information the accuracy of the solution will be degraded. Whether it’s degraded sufficiently to cause problems depends on how stable and accurate the other sources of information are on their own.
Sure, all of that applies if you are trying to derive attitude data from the gyros and other sensors.
But the yaw damper doesn’t care about the aircraft’s attitude in space. It only cares about yaw rate, plus lateral acceleration if it wants to keep the ball in the middle as well as doing the damping part.
tomjnx wrote:
You need multiple (2) antennas. You could compute the positions of the two antennae, and if you know where they’re mounted on your rigid body, that gives you the attitude of the body. If you do it a bit more cleverly than just computing two coordinates, most of the error sources cancel out, so it can be fairly accurate.
O boy. I didn’t expect that. That is pretty high accuracy…
tomjnx wrote:
And given that the wavelength of an L1 GPS signal is about 20cm, they don’t even need to be far apart, if your receiver can do carrier phase correlations.
Aha! Clever.
