AHRS tried to kill me

[PT20J]

I recently practiced using the Foreflight synthetic vision display on my iPad (Bluetooth linked to a Garmin GTX 345) as a backup in case my Aspen PFD failed. I quickly lost control of the airplane. Fortunately it was VMC and I just as quickly recovered. It turns out after some investigation that the attitude part of the AHRS in the less-than-a-year-old Garmin transponder had failed (the heading part works fine). I have had mechanical gyros fail and the failure is pretty evident because, after various erroneous indications, eventually they just stop indicating. But, I learned from this that the AHRS can fail in a way that it appears to be working -- it's just not displaying anything like the correct pitch or bank angles. 

I took some video of the AHRS indications in a standard rate turn to the left and a standard rate turn to the right. The iPad display is Garmin Pilot, but Foreflight does the same thing with one important difference: Foreflight never flags whereas apparently Pilot figures out that the AHRS input doesn't make sense and flags DEGRADED as shown in the video. 

So the moral to the story is that digital devices can fail, their failure modes are not simple or obvious, and they will kill you unless you quickly recognize the failure, and you need a backup that is in your scan to detect the failure. Garmin tells me that the AHRS used in the GTX 345 is the same as used in their primary flight displays.

Skip

[EricJ]

That's a bummer.   I've not tried a similar experiment with the cheapie AHRS in my Stratux, but I may try and do that.   I've always wondered why they need GPS to keep it aligned in the flight instruments, as that's seemed odd to me given it's application.

Edited February 27, 2020 by EricJ

[carusoam]

For best viewing... turn the iPad sideways to see both displays concurrently...

 

So... Skip,   
 

You wanted to see what would happen if you Actually followed the errant AHRS/iPad?

The resulting loss of control is pretty severe...   That is a bit disheartening...

We have a Garmin guy... @TrekLawler he can be found at BT...

Best regards,

-a-

[Yetti]

Why is garmin stuff failing so soon?

[PT20J]

1 hour ago, Yetti said:

Why is garmin stuff failing so soon?

Well, at least it was still under warranty. Garmin is sending a replacement unit to my avionics shop. 

[Ibra]

Thanks for the video, failure modes on tablet products are more along "NO connection", "NO GPS", "NO Battery", "New iOS update", anything else is just working fine 

6 hours ago, PT20J said:

Foreflight never flags whereas apparently Pilot figures out that the AHRS input doesn't make sense and flags DEGRADED as shown in the video. 

I think FF just display what they get as raw attitude/heading inputs, I am not sure if they even feed any annunciations or error signals?

But surely FF can do some pre-validation or integrity checks on what is displayed: a right bank on AI with ball dead in middle while HSI is turning left is not something "ready to display" without an FF warning? 

 

Edited February 27, 2020 by Ibra

[Guest]

I hope that Garmin uses better parts in their G 5’s.

Clarence

[tigers2007]

Dynon D3 Pocket Panel time! It’s nice having a totally independent backup hot-and-ready.
I’ve noticed some folks insist they don’t need a dedicated backup AI up and running; “I’ll pull it out if the AI fails”. Seems like when the AI fails they may only have a few seconds versus a minute or two to deploy a backup solution. (iPhone, iPad, etc). I’m glad you pointed this out as your ForeFlight display IS an “independent” backup for your AI and it failed.
Service Bulletin here we come!
I see their importance and possibly why some EFIS panel manufactures have the option for multiple ADHRS.

[jlunseth]

I don’t even turn my Foreflight AHRS display on. People were thinking early on that Foreflight plus a Stratus would be a cool attitude backup.  Well, it might be if the Stratus were immovably fixed to the frame of the aircraft, but if it moves at all the display is wrong, and AI is a display that can’t ever be wrong or if it is, you have to know immediately.  Here is another reason why.  So the question is, since it is unreliable and unreliable is unusable and deadly, why is it there to begin with? Need to have built in integrity checking or just not do it.

[Jerry 5TJ]

I’ve occasionally used the AHRS in the Garmin FlightStream 210 as a third AI for cross-check during flight.  The three sources have always agreed closely.

My FlightStream 210 is permanently mounted to the radio rack so the AHRS airframe reference is fixed:   Yes, you don’t want an AI linked to a box that’s moving around inside the plane.  

I don’t know if there is any internal self-check of the 210’s AHRS validity (of its ‘Integrity’ to use FAA-speak). Probably not.  

[Yetti]

10 hours ago, PT20J said:

Well, at least it was still under warranty. Garmin is sending a replacement unit to my avionics shop. 

    One of the design things that Netflix started was "design to fail"  The theory being.   Things are going to fail so just plan on it.   I think Dynon with the totally separate D10A running is a pretty good redundancy.     So if you are running a 10" screen with primary AHrs module,  a 7 inch screen with back up AHRS module.   Then the D10A  there are three AIs working for you. 

[midlifeflyer]

1 hour ago, jlunseth said:

I don’t even turn my Foreflight AHRS display on. People were thinking early on that Foreflight plus a Stratus would be a cool attitude backup.  Well, it might be if the Stratus were immovably fixed to the frame of the aircraft, but if it moves at all the display is wrong, and AI is a display that can’t ever be wrong or if it is, you have to know immediately.  Here is another reason why.  So the question is, since it is unreliable and unreliable is unusable and deadly, why is it there to begin with? Need to have built in integrity checking or just not do it.

I agree with your reasoning when using a portable AHRS source. OTOH, I really liked it when ForeFlight added the panel display - even without AHRS. Aside from the fact that it is an emergency, my partial panel issues have primarily been about being mental math challenged. On a heading of 290 and asked to turn to 010, I had to think too much to figure out it was 80 degrees (yes, I just used a calculator to make sure) and then divide by 3 to get a timed turn estimate. When the little Garmin 196-496 handhelds came along, their HSI page made a world of difference. I remember the first time I played with it. My safety pilot friend, not realizing what I was doing,  remarked he had never seen me fly partial panel that well. I pointed to the HSI page. Even that minuscule assist, by removing one, albeit simple, mental task from the "navigate" equation, allowed me to concentrate on "aviate."

[PT20J]

3 hours ago, M20Doc said:

I hope that Garmin uses better parts in their G 5’s.

Clarence

Well, that was really my point. They told me that this is the same AHRS they use in lots of stuff. A G5 might have similar failure modes. 
Skip

[steingar]

My iPad never tried to kill me, but my autopilot did.  Damn near succeeded, too.

[chriscalandro]

Has anyone else tried this with the same setup?  That seems like a very odd failure mode. 

It starts off what looks to be correct and then through the turn it slowly inverts the bank. 

At 180 degrees is it a perfect mirror?

[mooniac15u]

12 hours ago, EricJ said:

That's a bummer.   I've not tried a similar experiment with the cheapie AHRS in my Stratux, but I may try and do that.   I've always wondered why they need GPS to keep it aligned in the flight instruments, as that's seemed odd to me given it's application.

In the Stratux they use a MEMS gyroscope which is vibrational and tends to drift.  They use the GPS to compensate.  The issue is most significant in prolonged turns where the MEMS gyro loses track of the turn and starts to indicate level flight.  The developers worked on this problem for a long time before settling on using the GPS as a secondary information source.

[RobertGary1]

56 minutes ago, chriscalandro said:

Has anyone else tried this with the same setup?  That seems like a very odd failure mode. 

It starts off what looks to be correct and then through the turn it slowly inverts the bank. 

At 180 degrees is it a perfect mirror?

I was also wondering if it was actually using the Garmon ahrs or just the accelerometers in the iPad. Does FF have an indictor to show which it’s using? 

My shop has been installing G3 at a rapid pace in planes since it’s been out and not had a single failure. Of course most everyone buys two. 

 

-Robert 

[Ragsf15e]

8 minutes ago, RobertGary1 said:

I was also wondering if it was actually using the Garmon ahrs or just the accelerometers in the iPad. Does FF have an indictor to show which it’s using? 

My shop has been installing G3 at a rapid pace in planes since it’s been out and not had a single failure. Of course most everyone buys two. 

 

-Robert 

The attitude function in FF only works with an outside ahrs attitude source.  It will give a giant red x without a connected ahrs.  So it’s not a function of working off the iPad accelerometers. 
 

Supposedly these ahrs have a mtbf of 10,000hrs or greater (which I have no source for).  I think we should all thank Skip for having his fail so soon which ensures everyone else’s will last 20,000 hours! 

[Mooneymuscle56m]

Similar experience with the GRS 77 AHRS in the G1000, in 13 years, 1300 hours the unit has been replaced twice.  And it drives everything, moving map, attitude, autopilot, flight computer, heading indicator.  When it failed on me, I was down to mag compass and the backup on the far side of the panel. The iPad was handy.

Edited February 27, 2020 by Mooneymuscle56m

[EricJ]

5 hours ago, mooniac15u said:

In the Stratux they use a MEMS gyroscope which is vibrational and tends to drift.  They use the GPS to compensate.  The issue is most significant in prolonged turns where the MEMS gyro loses track of the turn and starts to indicate level flight.  The developers worked on this problem for a long time before settling on using the GPS as a secondary information source.

Yeah, my stratux puts up a warning when the GPS is either not connected or doesn't have a fix.   I think the Garmin AHRS, or any electronic AHRS, is going to be MEMS-based, which is likely why they have the GPS dependencies.   It still seems counter-intuitive to me to use a position source to determine attitude, rather than, say, averaging the acceleration vector to estimate where vertical is, which is more or less what an attitude gyro does.

Anyway, it's a good thing to know the behaviors so that panel redundancy can be managed.   To me it reinforces the notion that two G5s is a pretty good solution that covers functionality as well as redundancy.

[RobertGary1]

16 hours ago, EricJ said:

.

Anyway, it's a good thing to know the behaviors so that panel redundancy can be managed.   To me it reinforces the notion that two G5s is a pretty good solution that covers functionality as well as redundancy.

Assuming the failure mode is a red X. But if it’s  like the OP shows here and one G5 shoes a turn to the right and the other a turn to the left what do you do??

 

 -Robert 

[EricJ]

9 minutes ago, RobertGary1 said:

Assuming the failure mode is a red X. But if it’s  like the OP shows here and one G5 shoes a turn to the right and the other a turn to the left what do you do??

 

 -Robert 

I still have my turn coordinator, so that'd break the tie.

Edit:  It's no different than having any two redundant AIs.

Edited February 28, 2020 by EricJ

[mooniac15u]

1 hour ago, RobertGary1 said:

Assuming the failure mode is a red X. But if it’s  like the OP shows here and one G5 shoes a turn to the right and the other a turn to the left what do you do??

 

 -Robert 

Look at your magnetic compass?

[RobertGary1]

2 hours ago, mooniac15u said:

Look at your magnetic compass?

Turning from the north the compass will initially Indicate a turn in the opposite direction. It’s really the reason we have gyro dg’s. 
 

-Robert  

[mooniac15u]

1 hour ago, RobertGary1 said:

Turning from the north the compass will initially Indicate a turn in the opposite direction. It’s really the reason we have gyro dg’s. 
 

-Robert  

Didn't you learn to work with your magnetic compass (including accounting for that) as part of your partial panel training?  There will be a lag and then it will indicate which direction you are turning.  If the point is trying to decide which gyro is providing you the correct information then you will have your answer fairly quickly and you can switch back to using the gyro with the mag compass as confirmation.

You can also get some indication of which way you are turning from any moving map display in the cockpit including on a tablet.  Any of these things should point you towards the gyro with the correct information.