Goodbye GFC500 Pitch Oscillatons

[RoundTwo]

I’m happy to report that enabling the alternate gain settings on the latest software version on my G3X has completely eliminated the porpoising I’ve been dealing with.

I took the plane to the avionics shop for my altimeter certification and asked them to use the alternate gain settings we’ve heard about. I was told that I already had the correct version on the G3X but the standard gains were enabled. They enabled the alternate gains settings through configuration mode and on the flight home, the pitch oscillations I had going over were gone. YMMV, but this was the ticket for me.

[EricJ]

FWIW, this is an AD now:    AD 2024-14-03.    Updating various software to specified versions achieves compliance.

 

[PeytonM]

AD is for runaway pitch trim. The FAA turned Gamin’s SB into an AD. It does NOT include optional gain setting to address oscillations.

[EricJ]

29 minutes ago, PeytonM said:

AD is for runaway pitch trim. The FAA turned Gamin’s SB into an AD. It does NOT include optional gain setting to address oscillations.

The AD applicability is:

This AD applies to all airplane models specified in Table 1 to paragraph (c) of this AD, certificated in any category, having a Garmin
GFC 500 Autopilot System that includes an optional GSA 28 pitch trim servo installed per Supplemental Type Certificate No.
SA01866WI using Master Drawing List 005-01264-00, Revisions 1 through 76.

Mooneys are included in the table.   It doesn't mention the optional gain settings, but the updates that bring you that option are essentially required by AD now, or will be in August, anyway. 

 

[PT20J]

The software updates for the servos required by the AD are separate from the gain settings that were always available. The updated settings are only currently approved by the STC for the J and K although I have heard of some successfully trying them on long bodies. The installation instructions say to only use them if the standard settings do not work. I think Garmin just threw in the towel - it couldn’t ever figure out why the standard settings work with most airplanes but not on some. Bonanzas and Cirrus have also had pitch oscillation issues on some airplanes, so it’s not just a Mooney thing. Apparently the autopilot uses attitude augmented by rate for better pitch response. The alternate settings dial back the rate component (or maybe eliminate it). Closed loop control systems can be complicated. 

[Pinecone]

FYI, it is not enabling alternate gains, it is setting them to different values.  And some are quite different.

I had issues with IAS climbs only.   My shop stepped me through setting the alternate gains over the phone to avoid having to make a trip.

 

[RoundTwo]

7 minutes ago, Pinecone said:

FYI, it is not enabling alternate gains, it is setting them to different values.  And some are quite different.

I had issues with IAS climbs only.   My shop stepped me through setting the alternate gains over the phone to avoid having to make a trip.

 

That is incorrect. I know that because I asked specifically. For a certificated airplane, it’s all locked down. All you can do is select the standard or alternate. However, if your unit is set up as Experimental, you probably have more options.

[Pinecone]

As I said, I ACTUALLY DID THE GAIN CHANGES.  My shops walked me through the changes over the phone/

You set the desired settings from the SB.  For the initial setup the shop sets them according to the STC.  Any other gain settings are not legal for Certificated aircraft.

What you are saying is that there are profiles in the system, which there are not.   The SB just came out a month or so ago.  If what you said was true, older systems would have no option for the new gains.   Also, there would need to be different software for each make/model airplane.

[N201MKTurbo]

I have been designing and implementing PID control systems for over 30 years. One thing I have found is it is rare that two or more identical systems will operate the same with the same gain settings. You can have a standard set of gains that you use as the starting point for each new system and they will probably work well enough, but if you want optimal performance, the systems must be individually tuned. 
 

Most precision motion systems these days come with “auto tune” features. The drill is you start with your initial gains, then run the auto tune routine. If it works satisfactorily then you are done. If not you need to tune it by hand. 
 

The bottom line is a predefined set of gains is rarely going to be optimal.

[RoundTwo]

22 minutes ago, Pinecone said:

As I said, I ACTUALLY DID THE GAIN CHANGES.  My shops walked me through the changes over the phone/

You set the desired settings from the SB.  For the initial setup the shop sets them according to the STC.  Any other gain settings are not legal for Certificated aircraft.

What you are saying is that there are profiles in the system, which there are not.   The SB just came out a month or so ago.  If what you said was true, older systems would have no option for the new gains.   Also, there would need to be different software for each make/model airplane.

That’s really interesting because that is totally opposite of what I was told was available. I’m glad it worked out for you in the end.

[PT20J]

The GFC 500 STC documentation includes a Gain Addendum specific to each aircraft. What Garmin did was update the Gain Addendum for the M20J/K to include optional gain settings. The note says “Use gains in “Optional Value” column if pitch or pitch trim is overactive. Must use all values in “Optional Values”, if using any.”

[MikeOH]

2 hours ago, PT20J said:

The GFC 500 STC documentation includes a Gain Addendum specific to each aircraft. What Garmin did was update the Gain Addendum for the M20J/K to include optional gain settings. The note says “Use gains in “Optional Value” column if pitch or pitch trim is overactive. Must use all values in “Optional Values”, if using any.”

If I'm reading that right, you are trading one set of canned gain settings (the original STC settings) for another set of CANNED gain settings (you MUST use ALL of the optional values).  This seems DUMB!

While I don't have N201MKTurbo's 30 years of experience, I have tuned a few PID systems and they often need to be UNIQUELY 'tweaked'; substituting another 'best guess' set of parameters instead the first guess seems pretty suboptimal to me. BWTHDIK?

[PT20J]

14 minutes ago, MikeOH said:

If I'm reading that right, you are trading one set of canned gain settings (the original STC settings) for another set of CANNED gain settings (you MUST use ALL of the optional values).  This seems DUMB!

While I don't have N201MKTurbo's 30 years of experience, I have tuned a few PID systems and they often need to be UNIQUELY 'tweaked'; substituting another 'best guess' set of parameters instead the first guess seems pretty suboptimal to me. BWTHDIK?

That is correct. I believe the KAP/KFC 150 autopilots had fixed gains unique to each airplane type set by resistor values in potted modules in the flight computer. 

I suspect the problem is less about gains and more about the fact that the autopilot uses pitch rate and acceleration measurements in addition to attitude probably in an attempt to get sprightlier response. But, that’s a guess - I didn’t design the thing. 

[Pinecone]

On 7/21/2024 at 9:45 AM, N201MKTurbo said:

I have been designing and implementing PID control systems for over 30 years. One thing I have found is it is rare that two or more identical systems will operate the same with the same gain settings. You can have a standard set of gains that you use as the starting point for each new system and they will probably work well enough, but if you want optimal performance, the systems must be individually tuned. 
 

Most precision motion systems these days come with “auto tune” features. The drill is you start with your initial gains, then run the auto tune routine. If it works satisfactorily then you are done. If not you need to tune it by hand. 
 

The bottom line is a predefined set of gains is rarely going to be optimal.

Agreed.

 

It never made any sense that Garmin had specific gains in the STC.  I would have thought they would have starting gains and an allowable range.  Other that maybe the glass planes, each one is an individual

[Pinecone]

On 7/21/2024 at 9:55 AM, RoundTwo said:

That’s really interesting because that is totally opposite of what I was told was available. I’m glad it worked out for you in the end.

As others have said, there are two sets of gains.  You use one set or the other, but they are not in the system, you have to set them individually.

There is nothing physically stopping someone from tweaking the gains slightly.  Other that compliance with the STC.  And who would check that?

[N201MKTurbo]

So, are the gain police inspecting everybody‘s airplanes? Would the average IA even know how to check? Is that inspection on anybody’s checklist? 

[PT20J]

The gain table has 76 entries. The flight test procedure to determine settings is in the experimental G3X installation manual. 

 

[N201MKTurbo]

13 minutes ago, PT20J said:

The gain table has 76 entries. The flight test procedure to determine settings is in the experimental G3X installation manual. 

 

I haven't looked at the manual. Does it do some gain scheduling vs speed or something?

 

[MikeOH]

39 minutes ago, PT20J said:

The gain table has 76 entries. The flight test procedure to determine settings is in the experimental G3X installation manual. 

 

76???!!!  Are you frickin' kidding me?  Good grief, the odds of having all 76 set optimally, regardless of 'procedure' have to vanishingly small.  I can't fathom anyone designing an AP system with that many parameters.

I'm getting my STEC-30 repaired (if it ever fails) vs. going to that abomination of a design!

[BlueSky247]

Well, I imagine most of those settings should have some sort of descriptor or information to help guide the installer. If someone were determined, it should be easy enough to compare the two sets of values in the STC and what folks are using on the experimental side to tailor one's own setup. If so inclined...

[PT20J]

All the settings are not gains; some enable or disable optional items like trim servo and yaw damper,  set the direction of rotation of the servos for different installations, etc.

For pitch, the settings are max torque, servo gain, servo direction, clutch monitor enable, AP disconnect input polarity, min airpseed limit, max airspeed limit, vertical speed gain, vertical acceleration gain, airspeed gain, servo control mode, correction time, fine adjust amount, fine adjust time, vertical speed inertial gain, airspeed acceleration gain, airspeed tracking gain, airspeed gain schedule, altitude gain schedule. The bold ones are the ones that have optional settings. 

I think the reason for so much flexibility is that the GFC 500 was originally designed for experimental airplanes and there is probably a much wider range of flight characteristics in experimentals than certified airplanes that have to meet certifications standards.

[EricJ]

35 minutes ago, PT20J said:

All the settings are not gains; some enable or disable optional items like trim servo and yaw damper, servo gain set the direction of rotation of the servos for different installations.

For pitch, the settings are max torque, servo gain, servo direction, clutch monitor enable, AP disconnect input polarity, min airpseed limit, max airspeed limit, vertical speed gain, vertical acceleration gain, airspeed gain, servo control mode, correction time, fine adjust amount, fine adjust time, vertical speed inertial gain, airspeed acceleration gain, airspeed tracking gain, airspeed gain schedule, altitude gain schedule. The bold ones are the ones that have optional settings. 

I think the reason for so much flexibility is that the GFC 500 was originally designed for experimental airplanes and there is probably a much wider range of flight characteristics in experimentals than certified airplanes that have to meet certifications standards.

That's more complex than a tradition PID controller, so I suspect they got off in the weeds on a "make it work" expedition somewhere along the way during development and now they have something that is difficult to set up for a particular airplane application.   Or they just added a ton of knobs "because they can" in a software application.  On the old analog autopilots (like  Century) there are the usual expected adjustments and not much else.    When my Century III was working it'd fly an ILS like it was on rails, but it did have the usual autopilot oscillation during cruise.   Many do, as there's the usual complexity tradeoff.   I used to fly on corporate Embraer jets a lot and you could watch the wingtip trace a figure 8 constantly during cruise.   It really screwed with my OCD.  The pilots called it "the Embraer waltz".   I suspect that if Garmin added the additional complexity and still have oscillation issues they just got down a road where it grew out of control a bit and that's where it is.   The engineers in the group can probably relate.

 

[N201MKTurbo]

1 hour ago, PT20J said:

All the settings are not gains; some enable or disable optional items like trim servo and yaw damper,  set the direction of rotation of the servos for different installations, etc.

For pitch, the settings are max torque, servo gain, servo direction, clutch monitor enable, AP disconnect input polarity, min airpseed limit, max airspeed limit, vertical speed gain, vertical acceleration gain, airspeed gain, servo control mode, correction time, fine adjust amount, fine adjust time, vertical speed inertial gain, airspeed acceleration gain, airspeed tracking gain, airspeed gain schedule, altitude gain schedule. The bold ones are the ones that have optional settings. 

I think the reason for so much flexibility is that the GFC 500 was originally designed for experimental airplanes and there is probably a much wider range of flight characteristics in experimentals than certified airplanes that have to meet certifications standards.

I would imagine there are different control loops for different modes of operation. I doubt they are all in effect at the same time. It depends on if you are holding an altitude, holding an airspeed, or holding a vertical speed. If the pitch is unstable, you have to say which mode you are in when it is unstable.

And remember the same control loop needs to hold an altitude in smooth air and the same control loop needs to correct for severe turbulence. The two tasks are quite different, yet the same control system has to do both.

[Carl Everitt]

Chuck where did you find these gain settings ?

[RoundTwo]

13 minutes ago, Carl Everitt said:

Chuck where did you find these gain settings ?

I didn’t. My changes were done by my avionics shop. Maybe @Pinecone can help.