No idea but it seems to be totally solid.
Fairly obviously it does sense the gravity vector because it shows “level” when you power it up on the bench.
If there was a (certifiable) way of using a solid state gravity sensor, neither Garmin, Avidyne nor Aspen would bother taking all the external data like GPS and pitot into account.
I think reliability is read differently by all who replied. So the question should be more specific.
Be careful! Insisting on clear wording and exact vocabulary tends to be taken as pedantry (or worse) by some participants!
(for those who want them: 
)
But it could well be the opening question was meant about accuracy, perhaps dynamic or perhaps static or both, rather than reliability. Or perhaps “dependabability”.
achimha wrote:
If there was a (certifiable) way of using a solid state gravity sensor, neither Garmin, Avidyne nor Aspen would bother taking all the external data like GPS and pitot into account.
Maybe this is specific for MGL, I don’t know, but they use 3 rate gyros and 3 accelerometers (in 3 directions). I don’t understand how you can use a rate gyro without accelerometers (a “gravity sensor”). Does a “real” MEMS gyro even exist?
Peter wrote:
Fairly obviously it does sense the gravity vector because it shows “level” when you power it up on the bench.
Did you try powering it up at an angle? If it’s sitting level on the bench, it just might be assuming level when you power it up. But if you power it up tilted at 30 degrees, and it shows this correctly, it’s obviously doing something to sense gravity on startup.
Jesse wrote:
They don’t like rapid changes. They are generally not good for aerobatics etc.
That surprises me. These MEMS sensors are used in flybarless systems for RC helicopters. They cope with some pretty extreme aerobatics from the 3D guys.
Did you try powering it up at an angle?
I can’t test it now but I am 99% sure I did that and yes it did sense gravity.
achimha wrote:
If there was a (certifiable) way of using a solid state gravity sensor, neither Garmin, Avidyne nor Aspen would bother taking all the external data like GPS and pitot into account.
Huh? Why then do those certified backup attitude indicators not need GPS or airspeed input?
tomjnx wrote:
Huh? Why then do those certified backup attitude indicators not need GPS or airspeed input?
My question is rather why the Garmin etc. equipment do need GPS or airspeed input!
A mechanical AI maintains its attitude relative to the horizon using a free gyro. It has to compensate for gyro drift and does this by very slowly aligning the gyro with the gravity vector when the aircraft is flying (essentially) straight and level.
A solid-state gyro maintains a “virtual” attitude by integrating the output from rotation rate sensors. It also has to compensate for drift and could do this in the same way as a mechanical AI using gravity sensors.
So what are the position, heading, airspeed etc. used for?
