achimha wrote:
I hope you agree that should the electrical system in a Piper M600 fail in icing conditions, the only AI backup (Aspen EFD) would crap out within a minute or two.
I don’t know that particular setup but it seems problematic I agree. I am just concerned with a general no glass bakup view which is not valid. It is entirely installation and model dependent.
I have always had an electrically powered mechanical backup so am not trying to justify a particular view. But some installations are more robust than others.
JasonC wrote:
I am just concerned with a general no glass bakup view which is not valid. It is entirely installation and model dependent. […] But some installations are more robust than others.
And this thread has shown that pilots are not aware of the details which can be rather complex. Not even manufacturers seem to know it.
JasonC wrote:
I have always had an electrically powered mechanical backup so am not trying to justify a particular view.
Good thing is with this setup, there is very little you have to know and there are no hidden gremlins. That’s a big advantage in my view. I am not against all-glass but when I started looking into what manufacturers currently offer (after having seen the Piper M600 and the Aspen failure in that N210EU around the world video), my conclusion was that the move from a electrically powered gyro to all-glass backup instruments most likely reduces redundancy/reliability. At least in the M600, this was not well thought through by Piper. The MD302 used by TBM requires more information although I am quite sure it has similar issues.
achimha wrote:
However, it suggests that even the G5 uses both GPS and air data for erection.
G5 is certified for continuous flight in IFR as a primary in my plane, in the case GPS and Pitot static input fail. I use it as a backup though. With an internal backup battery of over 4 hours, its a reliable system. The two failure modes that still remain are failure of the LCD screen and software bugs. I had a many email exchanges when I was hit with software bugs. I asked the Garmin tech to explain the G5 internals. He was only willing to disclose that it uses a new combination of different types of sensors, one of which is a hall effect sensor. I asked whether the G5 uses Kalman filter for error correction, and surprisingly he said that they don’t. However, there are other error correction algorithm like Markov-chain Monte Carlo, Weiner filters, Bayes Estimators, so this may not mean much. Still to correct accelerometer errors Kalman filters are the obvious choice. In other words, the G5 uses some novel internal secret sauce, that makes it more sophisticated than the Aspen AI solution, and many others.
Kalman or Marcov chains is hardly relevant I think, that only is the method how external data is fed into the system.
The real question is — which sensor data does the G5 use to erect the AI? If it’s not pitot and GPS, what is it? A hall sensor measures current going through a wire, how would that apply to an AI?
Lucius wrote:
the G5 uses some novel internal secret sauce
I’m not sure the “secret sauce” answer is appropriate when trying to evaluate an instrument for suitability… All the constraints and possible failure modes need to be properly documented.
AF 447 had working attitude indication throughout, the only thing they lost was airspeed, and even that only for 30 seconds (primary) to 60 seconds (standby)
Aviathor wrote:
the GNS430s (…) do not provide GS
That must be a typo. Your GPS unit will absolutely provide ground speed :)
lionel wrote:
That must be a typo. Your GPS unit will absolutely provide ground speed :)
Sorry. No typo. In my mind GS was Glideslope. 
achimha wrote:
All the constraints and possible failure modes need to be properly documented.
Failure modes of the G5 are documented, and like I wrote, the failure of pitot static system and GPS will only degrade the AI solution slightly. Continuous flight in IMC is allowed under this failure mode. The internal secret sauce does not need to documented to the public.
Below is follow up answer received from MCIA, maker of MD302.
Hi Tomek:
Please see the answer from our Eng. Dept. below:
The ‘erection mechanism’ is based on a software algorithm associated with the solid-state gyros and accelerometers. However, the attitude solution does also rely on airspeed assistance to provide improved accuracy and performance. If airspeed is removed (plugged pitot port), both the airspeed and the attitude will ‘red X’ in order to avoid presenting potentially misleading information.
I apologize for the earlier misinformation.
Thank you for your continued support of MCIA. Have a great rest of your day.
Sincerely,
Wil Guy
Mid-Continent Instruments and Avionics
