Garmin dual attitude indicator failures with data

[Ragsf15e]

30 minutes ago, RobertGary1 said:

At least on the gi-275 it sits right on the glareshield like a handheld. I believe the g5 offers the same but I’ve not seen it. 

It is called a “glare shield antenna” in the certified g5 install guide.  In the experimental installation manual they give some basic antenna specs and tell you to use what works...

[Ragsf15e]

1 hour ago, RobertGary1 said:

There is speculation that my dual AHRS failure was a result of a gps feed issue. Also if you lose ship power and lose both gps and pitot (icing) you’d lose all AHRS too. Garmin suggested adding local antenna as it will runoff the display battery.  So it’s redundancy  for the AHRS attitude 

 

The AHRS requires external feeds to slew such as pitot or gps. 

Do they know how a gps feed issue caused the airspeed to fail too?  Isn’t that straight from the pitot?

[jetdriven]

I find it disturbing that loss of pitot input and loss of GPS data causes the AHRS to outright fail. That stuff should be bulletproof because they tell you you can remove everything else on the airplane but those two devices

[PT20J]

The G5 requires GPS for AHRS aiding and other functions. You can use the internal receiver with the glareshield antenna, or an RS-232 connection to a GPS navigator such as a GNS or GTN. During configuration, you set whether to enable the internal receiver or use the external source. From the documentation it does not appear that there is any automatic failover. The internal antenna isn’t mentioned in the certified version documentation - probably because it doesn’t work very well as noted above. I cannot find a mention in the G5 Installation Manual of an external antenna mounted to the fuselage, but the G3X Installation Manual mentions that if a G5 is already installed and is to be retained as a backup and it has an external antenna then the external antenna must be disconnected though it may be used for another unrelated GPS. So, this implies that there are approved G5 installations with fuselage mounted antennas.

The glareshield antenna can be bolted to the glareshield or attached with Velcro.

Skip

[PT20J]

19 minutes ago, Ragsf15e said:

Do they know how a gps feed issue caused the airspeed to fail too?  Isn’t that straight from the pitot?

The OP reported that his pitot has one of those flip up covers and it didn't flip up. The interesting thing is why that caused the AHRS to tank since it should get velocity input from the GPS to correct it. It appears that there was something wonky with the GPS as well.

[aviatoreb]

1 hour ago, RobertGary1 said:

At least on the gi-275 it sits right on the glareshield like a handheld. I believe the g5 offers the same but I’ve not seen it. 

I don’t see what item it would be - do you know?

[201Mooniac]

As per Steve from Garmin the G5 will use the GPS with the best accuracy so if you connect both a GTN/GNS and the glareshield antenna it will use whichever it decides is best.  Therefore if one fails (like the GTN/GNS will if you lose power) then the other will be the "best accuracy".  I installed the glareshield antenna for both my G5 and my TXi.  Probably overkill but I like any redundancy I can get.

[PT20J]

I've been reading a lot of manuals and ferreting out information from Garmin Support posts on Beechtalk and other sources because I'm about to go with a G3X/G5 install and ditch the vacuum system and so my understanding of Garmin AHRS is important to me. I'm trying to improve reliability and safety, not reduce it. My only experience with Garmin AHRS is GTX 345 which suffered a AHRS failure after a year which Garmin replaced it on warranty with no hassle. But during the troubleshooting, Garmin Support told me it has the same AHRS as other Garmin products which has me wondering. Understanding the AHRS  isn't easy because Garmin doesn't publish a lot of details about how its products actually work. 

Anyway, MEMS based AHRS usually require some additional aiding to compensate for drift and acceleration errors. All AHRS have three axis accelerometers and gyros. Some (as used in the RCA2610) also have three axis magnetometers. With the magnetometers, the errors are much less and this is why the RCA2610 doesn't need aiding according to the person I spoke with at RC Allen. Garmin doesn't mention magnetometers in the AHRS so it appears they don't have them and instead use aiding. I would expect Garmin AHRS with aiding to be more accurate than the RCA2610 when the aiding works and less accurate when it does not.

If the GI 275 works like the G5 the aiding comes from the internal GPS if connected to an antenna and enabled. If that GPS is bad or unavailable, it uses a MapMX GPS input from an external GPS navigator (assuming one is wired in). If both GPS sources are bad or unavailable, it uses pitot/static. So the most robust system would have an antenna connected to the unit and the internal GPS enabled, so that two GPS sources have to be bad before it goes to pitot/static aiding.

Garmin Support says that the AHRS meets all the required maneuvers in the degraded (unaided) requirement of RTCA DO-334 over a two hour period.

So how can this thing screw up? Well, here's one guess. It's easy to tell if a GPS is completely dead, in which case the software can switch to either the other GPS source or pitot/static for aiding. But, if a GPS source is just putting out some bad data, it might be difficult for the software to determine this, and it might be possible for the bad data to screw up the AHRS. Without knowing the design details, it's impossible to know. But here is an interesting thread.

https://www.beechtalk.com/forums/viewtopic.php?p=2153712#p2153712

Skip

[Yetti]

On 5/24/2021 at 5:44 PM, RobertGary1 said:

There is speculation that my dual AHRS failure was a result of a gps feed issue. Also if you lose ship power and lose both gps and pitot (icing) you’d lose all AHRS too. Garmin suggested adding local antenna as it will runoff the display battery.  So it’s redundancy  for the AHRS attitude 

 

The AHRS requires external feeds to slew such as pitot or gps. 

That speculation is not supported by the data file you posted.   It is saying 3d GPS fix and Lat Long the whole length of the file.  Including the times when the attitude information cuts out.   They need to dig deeper.

[Yetti]

@RobertGary1   Did you get the file from the other unit?

[PT20J]

It's interesting for the engineers among us to try to figure this stuff out. But in the end we can never fully understand it because the details of the hardware and the software are unknown to us. It's Garmin's job to continually improve the products while dealing with new failure modes as they are discovered. I have no doubt whatsoever that Garmin has made  every effort to make their devices fault tolerant through design and rigorous testing. However, having spent much of my career in product development of embedded systems, I know that many interesting things happen after a product is released into the the real world.

The more important question is how does the owner/operator protect against rare but catastrophic failure modes. The standard method requires some sort of technology diversity -- i.e., not depending on a single technology. If Garmin's statement to me was correct and the same AHRS technology is used in all current products, then it is possible that a failure mode might present itself in multiple products. This would amount to being dependent on a single technology if the primary and backup system are both made by the same manufacturer. So there are two obvious approaches to increasing safety: 1) retain the vacuum system and mechanical attitude indicator, and 2) add a electronic backup attitude indicator made by a different manufacturer using a different technology. Since in my case, the G3X STC requires a G5 backup, and there are system integration reasons why this is a desirable configuration, this would mean adding a third attitude indicator. 

Skip

 

[DCarlton]

16 minutes ago, PT20J said:

It's interesting for the engineers among us to try to figure this stuff out. But in the end we can never fully understand it because the details of the hardware and the software are unknown to us. It's Garmin's job to continually improve the products while dealing with new failure modes as they are discovered. I have no doubt whatsoever that Garmin has made  every effort to make their devices fault tolerant through design and rigorous testing. However, having spent much of my career in product development of embedded systems, I know that many interesting things happen after a product is released into the the real world.

The more important question is how does the owner/operator protect against rare but catastrophic failure modes. The standard method requires some sort of technology diversity -- i.e., not depending on a single technology. If Garmin's statement to me was correct and the same AHRS technology is used in all current products, then it is possible that a failure mode might present itself in multiple products. This would amount to being dependent on a single technology if the primary and backup system are both made by the same manufacturer. So there are two obvious approaches to increasing safety: 1) retain the vacuum system and mechanical attitude indicator, and 2) add a electronic backup attitude indicator made by a different manufacturer using a different technology. Since in my case, the G3X STC requires a G5 backup, and there are system integration reasons why this is a desirable configuration, this would mean adding a third attitude indicator. 

Skip

 

The question I continue to ask though, is why do Garmin systems *require* a GPS input, and the Sandia, King and Uavionix AI replacements *do not*.  That's something we should all be able to understand.  The FAA should be able to explain it if no one else.  They reviewed and processed the test and certification.  And all of these systems have now apparently experienced some bugs.  There shouldn't be any deep design secrets if these units have gone through Govt. certification.  

Edited May 30, 2021 by DCarlton

[EricJ]

Mechanical attitude gyros "drift" or precess as well and use gravity to realign.  I'm curious about the physics of why a system with MEMS accelerometers and gyros needs GPS to work, but most say they do.

  

[philiplane]

When the G1000 first came out, it had lots of attitude failures due to miscompares between AHRS, GPS heading, and magnetic heading. Most of these occurred on the ground, during taxi. Fast turns, turns near large metal objects like storm drain covers, buried conduits, etc, would cause problems. So they've had problems with GPS aiding their AHRS from the beginning, this is just the same problem with the latest product. Software revisions fixed the earlier problems. But as the AHRS sensors got smaller, the sensitivity to rate changes increases. Compare a 2 inch peanut gyro to a 5 inch AN gyro, and you get the idea. I'm not a fan of all-electronic panels that rely upon GPS aiding. I'd rather keep a vacuum gyro with dual vacuum sources as a back up. In 30 years of flying, I've had one vacuum AI actually fail due to loss of a vacuum pump, and that was a wet pump that ingested a piece of debris after other system maintenance. In the last 5 years, I've had at least 24 glass panel failures in various glass planes for various reasons. Some had bad AHRS units, some displays failed, some ADC's failed, and so on. Garmin, Aspen, and Avidyne systems about equally represented.

Many years ago, humans designed attitude indicators and directional gyros. Each item had one job, so it's easy to make it do that job well. Today's humans are designing the glass systems that integrate everything, and it's hard to imagine every possible failure scenario. Especially when one person does the mechanical design, another does the interface details, another does the electrical, and another does the software. Each contribution is a potential point of failure if the final review doesn't ferret out the flaws. Fallibility is one of our prime problems as humans. Maybe we need primates to do this job?

Edited May 30, 2021 by philiplane

[PT20J]

1 hour ago, DCarlton said:

The question I continue to ask though, is why do Garmin systems *require* a GPS input, and the Sandia, King and Uavionix AI replacements *do not*.  That's something we should all be able to understand.  The FAA should be able to explain it if no one else.  They reviewed and processed the test and certification.  And all of these systems have now apparently experienced some bugs.  There shouldn't be any deep design secrets if these units have gone through Govt. certification.  

Well the Sandia and King (which is a rebranded Sandia) were the subject of an Emergency AD last year because of several failures. The AD limits their use to non-autopiltot coupled day VFR. I don't know if there was a fix to lift the AD or not.

Skip

[MikeOH]

WOW!  Reading through all of this, plus what I've read on other fora, I honestly don't see how one can believe that safety and reliability are being increased!  Seems to me the opposite!  Certainly mechanical gyros wear out...over a long period...and with plenty of warning in nearly all cases.  While the vacuum pump doesn't give warning, if you are proactive with replacement you can achieve practical reliability.

Contrast with EXTREMELY complex electronic systems dependent upon multi-sensor input integrity and literally millions of transistors, not to mention millions of lines of code.    What could possibly go wrong??

As @PT20J accurately points out, we are NEVER going to know how these systems work at any level of practical detail. What scares me, and short term history (who knows what long term will be!) shows, is that neither do the manufacturers

I'm not without career experience with INS systems which relay upon accelerometers, MEMS or otherwise.  You measure their error in miles per hour (makes sense if you think about it); stop to think about all the errors that MUST be ACCURATELY compensated for (temperature is a prime example) in order to achieve performance that does NOT require GPS, or other independent, position/heading information.

Despite all the wonders of digital electronics the world is still analog and the sensors between those two worlds is all important.  Throw in all the unanticipated software and software/hardware interactions, and I just don't believe present day electronic systems are an improvement in safety and reliability. Flame suit on!

I realize that, on the whole, pilots like new fancy gadgets and gizmos.  Just suggesting not to fall in love with new technology for its own sake, but for real improvement.

Signed, a 30 year cynical EE

[Ragsf15e]

Even with their issues, I definitely like my G5s over a vacuum adi plus T&B.  I feel like they have much less chance of failure and will “red x” if they do, so at least I know.  As it is, I could technically fly partial panel without the G5s because I had to keep all the other stuff, but I doubt that would be fun.  Also, the T&B doesn’t have battery backup, so it might be toast with the G5s depending on why they failed.

Where I agree with you, it would be nice to have a backup with different technology/software/aiding requirements.  Vacuum AI would be ok, but I think an RCA adi (which doesn’t require aiding as PT20j has explained) would be fine.  If My stec30 wouldn’t be part of the T&B, I’d probably put in the “3rd” adi.  So 2xG5s and 1 rca.  Probably still much lighter than vacuum ai.

Edited May 30, 2021 by Ragsf15e

[Yetti]

the track record is not good here.  2 vendors Aspen/Garmin with semi serious failure modes.   It does make me want to query Dynon about their processing paths.   That said I think it would have showed up in the experimental market.

Sad part is this is all 1990s technology.   The sadder part is some of the pipelines I have worked for are still running 1980s technology.

Edited May 31, 2021 by Yetti

[Andy95W]

We’re only hearing the negative information.  There have been a few AV30 problems, a handful of Aspen issues (which appear to be fixed), and 1 or 2 Garmin failures.  This particular Garmin problem is due to a rare double failure- loss of pitot and loss of GPS connection.

What we’re not hearing about are the thousands of trouble-free installations.

I’m very comfortable with my dual G5’s and no vacuum system, although I do have an STEC autopilot with its turn coordinator as my emergency backup.

[PT20J]

28 minutes ago, Andy95W said:

We’re only hearing the negative information.  There have been a few AV30 problems, a handful of Aspen issues (which appear to be fixed), and 1 or 2 Garmin failures.  This particular Garmin problem is due to a rare double failure- loss of pitot and loss of GPS connection.

What we’re not hearing about are the thousands of trouble-free installations.

I’m very comfortable with my dual G5’s and no vacuum system, although I do have an STEC autopilot with its turn coordinator as my emergency backup.

You make a good point. The troubling part for me though is that Garmin says their AHRS meets the RTCA DO-334 with no GPS or pitot static aiding for 2 hours. So even a double failure should not be catastrophic. Clearly something else happened here.

[MikeOH]

Just now, PT20J said:

You make a good point. The troubling part for me though is that Garmin says their AHRS meets the RTCA DO-334 with no GPS or pitot static aiding for 2 hours. So even a double failure should not be catastrophic. Clearly something else happened here.

Do you know what the allowable errors are after 2 hours per DO-334?  I have no familiarity with that specification/standard.

[MikeOH]

42 minutes ago, Andy95W said:

We’re only hearing the negative information.  There have been a few AV30 problems, a handful of Aspen issues (which appear to be fixed), and 1 or 2 Garmin failures.  This particular Garmin problem is due to a rare double failure- loss of pitot and loss of GPS connection.

What we’re not hearing about are the thousands of trouble-free installations.

I’m very comfortable with my dual G5’s and no vacuum system, although I do have an STEC autopilot with its turn coordinator as my emergency backup.

"a few" here, and "a handful" there...what does that add up to 10, 30, 50?  For products that are pretty new.  How many are out there and installed?  Call 'thousands" 5,000.  10 out of 5,000 is 2,000 ppm failure rate.  Not too impressive for something your life may depend upon, I'm afraid.

I'd bet (and, I'm trying to research) vacuum driven gyros are MUCH better than that.

[PT20J]

8 minutes ago, MikeOH said:

Do you know what the allowable errors are after 2 hours per DO-334?  I have no familiarity with that specification/standard.

No, I don't have a copy. But the RTCA has been setting the standards for this stuff for a long time and many FAA TSOs reference RTCA documents. You can buy a copy for $150 at https://www.rtca.org/ if you are really interested.

Skip

[DCarlton]

1 hour ago, Andy95W said:

We’re only hearing the negative information.  There have been a few AV30 problems, a handful of Aspen issues (which appear to be fixed), and 1 or 2 Garmin failures.  This particular Garmin problem is due to a rare double failure- loss of pitot and loss of GPS connection.

What we’re not hearing about are the thousands of trouble-free installations.

I’m very comfortable with my dual G5’s and no vacuum system, although I do have an STEC autopilot with its turn coordinator as my emergency backup.

Check out the emergency AD status of the Sandia SAI-340A and King KI-300.  

I'm planning to scale back my vacuum system but keep the EVT Turn Coordinator and associated DG.  

Edited May 31, 2021 by DCarlton

[PT20J]

7 minutes ago, DCarlton said:

Check out the emergency AD status of the Sandia SAI-340A and King KI-300.  

I'm planning to scale back my vacuum system but keep the EVT Turn Coordinator and associated DG.  

You still need a pitch reference. Remember, partial panel is needle, ball and airspeed. Altitude would be good, too.