GFC-500 Disconnecting

[Rick Junkin]

1 hour ago, PT20J said:

g3xpert mainly deals with the non-certified customers and the latest experimental version is 9.41 so perhaps this is something already addressed that hasn't made it to a certified release yet. 

...which begs the question again, can I load the release that has experimental v9.41 and select certified in the config and still get the correct certified v9.17, but perhaps with some bugs worked out? I know it wouldn't be legal, but would it work? 

[BlueSky247]

You fellas are way more patient than I am. I’d be livid had I coined for a gfc-500 install and Garmin was not actively troubleshooting the problem. 

[PT20J]

2 hours ago, Rick Junkin said:

...which begs the question again, can I load the release that has experimental v9.41 and select certified in the config and still get the correct certified v9.17, but perhaps with some bugs worked out? I know it wouldn't be legal, but would it work? 

Hard to say. From what I understand, there is one G3X Touch code base. So, the 9.XX versions would be sequential from 9.00 (9.01, 9.02... 9.41). Since Garmin skips version numbers in its releases, some are likely versions cut for internal testing but not released to the field. Also, some versions in the sequence are released for certified and some for noncertified. The most current certified version is 9.17. There is no noncertified 9.17. The closest noncertified released versions were 9.14 and 9.22. Garmin's release notes are not very detailed and often say something like "General improvements to system operation" which likely means minor bug fixes. So, it's unclear what you would get with 9.41 running in certified mode. The biggest problem I see is that I don't know how Garmin does it's updates. It could be a complete reload or more likely it is a patch. Sometimes it can be problematic to back out a patch to get back to an earlier version, especially if several versions were skipped.

[Pinecone]

17 hours ago, PT20J said:

Did the D99 happen coincident with the AP Fail? If so, perhaps the GSU 25D is the problem rather than the autopilot. Without understanding the design, it's not possible to know, but maybe the "system" referred to in the failure message is the ADAHRS and not the autopilot. I could see the autopilot perhaps wanting to disengage if the ADAHRS was degrading but perhaps not degraded enough to cause a MISCOMP with the standby. Or maybe it is something external that affects both ADAHRS.

Were the disconnects in different locations to rule out GPS interference in some location?

Nope, N100 N100 at the time of AP Fail.

[emes]

Curious if you ever solved this...

I've had the same/similar issue and there don't seem to be any consistent patterns in the diagnostic logs before the failure.  I have a combo of the -11 and -21 servos currently, with (2) GSU-25D and (2) GMU-11.  This has been happening across several software releases.  It will sometimes fail enroute, but usually fails when I need it most during approach in imc.

[Pinecone]

Yes, mine seems to be solved.

My shop suggested to go back the original STC gain settings, just to see what would happen.

When the checked the dealer site and got the latest STC gain settings, they saw that they had changed from the original settings.  They set the gain settings to the newer STC settings.  I had one disconnect right after doing this, and NONE since that one.  And over 30 hours on the new settings and everything is working great

[PT20J]

7 hours ago, emes said:

Curious if you ever solved this...

I've had the same/similar issue and there don't seem to be any consistent patterns in the diagnostic logs before the failure.  I have a combo of the -11 and -21 servos currently, with (2) GSU-25D and (2) GMU-11.  This has been happening across several software releases.  It will sometimes fail enroute, but usually fails when I need it most during approach in imc.

Is this an experimental installation? Only one GSU 25D is approved by the STC for certified installations. If it's experimental, I believe g3xpert will evaluate the assert logs and help you get to the bottom of it.

[emes]

3 hours ago, Pinecone said:

My shop suggested to go back the original STC gain settings, just to see what would happen.

Interesting.  I don't believe mine have changed but will confirm.  Also congrats.  Finger crossed.

2 hours ago, PT20J said:

Is this an experimental installation? Only one GSU 25D is approved by the STC for certified installations. If it's experimental, I believe g3xpert will evaluate the assert logs and help you get to the bottom of it.

Certified.  Hmm.

[Pinecone]

OK, I talked to my shop this past week.

When they installed my GFC-500, they set the gains to the current STC document.  As mentioned, I had pitch excursions.  So the gains were switched to the gains in the J/K service bulletin from Garmin.  This fixed the IAD climb pitch excursions that some people had.

I then was getting AP disconnects.  So one thing my shop wanted to try was going back to the STC settings.  A new Revision of the STC was issued in July 2024.  I picked up my plane in mid-June 2024, so the previous Revision.

The July 2024 version of the STC has different gain settings than the version before.  My shop installed these in September 2024

In my plane, the July 2024 gain settings do not exhibit any pitch hunting (maybe 1 knot or so) and no uncommanded disconnects.

So, if your gains were set before July 2024 in a J or K, and you are getting uncommanded disconnects, have your shop change them to the new gain settings in the July 2024 Revision of the STC.

[Ragsf15e]

On 3/31/2025 at 2:24 PM, Pinecone said:

OK, I talked to my shop this past week.

When they installed my GFC-500, they set the gains to the current STC document.  As mentioned, I had pitch excursions.  So the gains were switched to the gains in the J/K service bulletin from Garmin.  This fixed the IAD climb pitch excursions that some people had.

I then was getting AP disconnects.  So one thing my shop wanted to try was going back to the STC settings.  A new Revision of the STC was issued in July 2024.  I picked up my plane in mid-June 2024, so the previous Revision.

The July 2024 version of the STC has different gain settings than the version before.  My shop installed these in September 2024

In my plane, the July 2024 gain settings do not exhibit any pitch hunting (maybe 1 knot or so) and no uncommanded disconnects.

So, if your gains were set before July 2024 in a J or K, and you are getting uncommanded disconnects, have your shop change them to the new gain settings in the July 2024 Revision of the STC.

interesting.  I might have to look into that eventually.  I have about 25 hours in my new plane and it's had g3x/gfc500 for 2.5 years.  I had my first disconnect yesterday.  Looking at the logs, got the "Fail / System not useable" code.  Only thing I noticed was the YD was at like 33 torque for a while before the disconnect.  During the disconnect, the YD stayed engaged.  I manually disengaged, then re-engaged and the YD went down to like ~7 and everything worked fine for the rest of the flight.  Logs show good gps and attitude.

As this happened, I was on departure, full power, 120kias climb.  I probably had a weak right foot... could that be making the YD work harder and cause a disconnect?  And if so, why would the YD remain engaged?

@PT20J do you still get a random disconnect from time to time?

[PT20J]

25 minutes ago, Ragsf15e said:

 do you still get a random disconnect from time to time?

No, I haven't had a disconnect in a long time and I only had two. The first, there was an uncommanded roll and near simultaneous disconnect. The logs did not show that the servo caused the roll so it might have been turbulence overpowering the roll servo. I don't know and it never happened again.

The second was level in cruise in ALT on a bumpy day. It disconnected and I reengaged and it disconnected again. Subsequently, I initiated a climb to get out of the turbulence and reengaged during the climb in VS (I think) and it worked fine. The logs showed that it disconnected due to a miscompare between the G3X and GMC 507. It's been flawless for a couple of years.

Skip