On 12/4/2020 at 2:03 PM, Mikeingreen said:

Good luck in your search, Robert. I hope someone steps up.

Thank you. As it happens, someone did step up and it looks promising.

On 12/4/2020 at 12:20 PM, hvw said:

Robert,

I'm not a programmer, but I do wish you luck with your efforts to tackle the A/P heading issue. 

Hans

Thanks Hans.

6 minutes ago, robert young said:

Thank you. As it happens, someone did step up and it looks promising.

Robert,

Can you describe the approach you all are taking here?

-Matt

The AP did seem to work pretty much in standard fashion prior to the latest patch, so it wasn’t far removed from reality, other than some of the misbehavior like the “death spiral” some have reported, (which never happened to me, but don’t doubt it did affect some users), and various pitch or roll oscillations. For me, the main experience of roll oscillation has been in the Airbus. It was very bad in the Alpha, and much better in the release, but it still does it. I understand it affects the Longitude as well, but I have never flown that one in the sim as yet.

One operator I worked for a few years ago had a Gulfstream G200 business jet that developed the “dolphin dance” problem after coming out of a major inspection at a MRO (who shall remain unnamed). In level flight, the pitch would oscillate up and down continuously with altitude hold engaged. That turned out to be a mechanical problem. The tension on the control cables connecting the autopilot pitch servo to the elevator had been set much too low, causing free play.

When I first got the release version of the sim in August, I took the analog C172 up to test its flight model performance compared to the real aircraft, including stalls, slow flight etc. I did do left and right timed standard right turns based on the turn coordinator, and they were accurate almost to the second, so turn rate is definitely available as a sim variable.

That aircraft is encrypted, but the C152 is not. I’m curious if the code for the turn coordinator gauge is exposed to see what variables are driving it.

Edited December 5, 20205 yr by JRBarrett

This is a thread for flightsim geniuses haha!  You lost me at autopilot.

22 minutes ago, MattNischan said:

Robert,

Can you describe the approach you all are taking here?

-Matt

I'll PM you Matt.

7 minutes ago, JRBarrett said:

That turned out to be a mechanical problem. The tension on the control cables connecting the autopilot pitch servo to the elevator had been set much too low, causing free play.

I think they call that "non-linear system behavior". 😂

8 minutes ago, JRBarrett said:

I did do left and right timed standard right turns based on the turn coordinator, and they were accurate almost to the second, so turn rate is definitely available as a sim variable.

Yes indeed, the turn coordinator indication is available as a readable simvar (two really, one for turn rate and one for sideslip amount).

-Matt

17 minutes ago, MattNischan said:

I think they call that "non-linear system behavior". 😂

Yes indeed, the turn coordinator indication is available as a readable simvar (two really, one for turn rate and one for sideslip amount).

-Matt

For comparison, I just did an autopilot-controlled turn in X-Plane 11 in the analog 172. Altitude 3500 feet, speed 100 KIAS. Using heading select and altitude hold, I commanded a 180 degree turn with the heading bug. The AP rolled directly to standard rate per the turn coordinator, and held it throughout the turn. Roll out began 12 degrees before the selected heading. The resulting bank angle was about 15 degrees, and it took 65 seconds.

I then firewalled the throttle at let the aircraft accelerate to 128 knots. A 180 degree turn was again done at standard rate per the turn coordinator, and again took 65 seconds. The only difference was that bank angle was 20 degrees at the higher airspeed.

This particular autopilot emulates an STEC 55X, which is strictly rate-based for lateral steering.

3 minutes ago, JRBarrett said:

For comparison, I just did an autopilot-controlled turn in X-Plane 11 in the analog 172. Altitude 3500 feet, speed 100 KIAS. Using heading select and altitude hold, I commanded a 180 degree turn with the heading bug. The AP rolled directly to standard rate per the turn coordinator, and held it throughout the turn. Roll out began 12 degrees before the selected heading. The resulting bank angle was about 15 degrees, and it took 65 seconds.

I then firewalled the throttle at let the aircraft accelerate to 128 knots. A 180 degree turn was again done at standard rate per the turn coordinator, and again took 65 seconds. The only difference was that bank angle was 20 degrees at the higher airspeed.

This particular autopilot emulates an STEC 55X, which is strictly rate-based for lateral steering.

Now that is interesting having flown both the STEC and Garmin A/P's and servos. Shouldn't these be digital and not rate based on an old STEC? Does this hold true for the other planes or just the 172? Asking for a friend 🙂

1 hour ago, Flyfaster_MTN002 said:

Now that is interesting having flown both the STEC and Garmin A/P's and servos. Shouldn't these be digital and not rate based on an old STEC? Does this hold true for the other planes or just the 172? Asking for a friend 🙂

It doesn’t matter if the servos, air data and pitch/roll information, (and the autopilot controller itself) are digital or analog, or a combination of both. The physics involved are the same. Heading changes are always based on rate of turn, (not bank angle), unless the airspeed is high enough that bank angle is the limiting factor, it which case the autopilot will command the maximum allowable steady-state bank angle for the current airspeed when executing a heading change.  Of course, bank angle will vary as the the aircraft rolls into and out of a turn, and can vary in the turn itself when trying to maintain a specific ground track in Nav mode, but in a pure heading mode turn of more than a few degrees, the autopilot will always use turn rate as the state control.

With an older S-TEC autopilot such as the 55X, when installed in a “steam gauge” cockpit, the existing turn coordinator will be replaced by an S-TEC turn coordinator which provides the same visual cues for the pilot as the original unit, as well as outputting turn rate information to the autopilot controller.

A newer S-TEC system like the 3100, everything is digital end-to-end.

In my current job, I work on both CRJ-200 and Dassault Falcon 900EASy aircraft. The 1990s-vintage Collins FCC-4010 autopilot on the CRJ is a hybrid system. Most system inputs are digital, but the actual pitch and roll steering and yaw damping is done via analog servo loops. The newer Honeywell autopilot in the Falcon actuates the servos and yaw damper via a digital CAN bus.

MSFS emulates a variety of autopilot types, from simple to complex, but observing AP behavior in a basic heading mode turn is a good way to see what might be going on “under the hood”, and after the latest patch, it’s obvious that there is something wrong with the control logic. Hopefully the next update later this month will correct that.

 

Edited December 6, 20205 yr by JRBarrett

55 minutes ago, JRBarrett said:

MSFS emulates a variety of autopilot types, from simple to complex, but observing AP behavior in a basic heading mode turn is a good way to see what might be going on “under the hood”, and after the latest patch, it’s obvious that there is something wrong with the control logic. Hopefully the next update later this month will correct that.

 

Ok so no Autoland anytime soon. Got it 🙂

Have flown behind the Sorcer, Stec-55X, Avidyne DFC-90 and Garmin. From worst to best the Sorcer is basically trying to kill you, and the Garmin is on rails compared to the Stec. This last paragraph was the crux of what I was getting at thanks!

10 hours ago, JRBarrett said:

I do have many years of technical experience maintaining and installing real autopilot systems, ranging from small GA to transport category aircraft.

I am just wondering, on G1000 equipped aircrafts with GFC700 autopilot, do they use specific AP "code, logic, servo value" for each plane or it's universal?

9 hours ago, JRBarrett said:

The AP did seem to work pretty much in standard fashion prior to the latest patch, so it wasn’t far removed from reality, other than some of the misbehavior like the “death spiral” some have reported, (which never happened to me, but don’t doubt it did affect some users), and various pitch or roll oscillations. For me, the main experience of roll oscillation has been in the Airbus. It was very bad in the Alpha, and much better in the release, but it still does it. I understand it affects the Longitude as well, but I have never flown that one in the sim as yet.

One operator I worked for a few years ago had a Gulfstream G200 business jet that developed the “dolphin dance” problem after coming out of a major inspection at a MRO (who shall remain unnamed). In level flight, the pitch would oscillate up and down continuously with altitude hold engaged. That turned out to be a mechanical problem. The tension on the control cables connecting the autopilot pitch servo to the elevator had been set much too low, causing free play.

When I first got the release version of the sim in August, I took the analog C172 up to test its flight model performance compared to the real aircraft, including stalls, slow flight etc. I did do left and right timed standard right turns based on the turn coordinator, and they were accurate almost to the second, so turn rate is definitely available as a sim variable.

That aircraft is encrypted, but the C152 is not. I’m curious if the code for the turn coordinator gauge is exposed to see what variables are driving it.

You can't fix the current broken heading Asobo AP behavior of "rolling to wings level starting 30 degrees away from the target" with any accessible variable in the .cfg files or SimConnect variable or variable accessible via C++.  The Asobo AP SimVars accessible via SimConnect or C++ are all read-only values, and the aircraft .cfg settings for PID don't fix the behavior due to inaccessible changes to the AP internal code which Matt described earlier in this thread.  The updated PID values from Seb fix a separate issue where some aircraft could only bank in one direction.

Of course one could could a complete AP replacement using their own code for the AP, their own cockpit button/glass cockpit code updates to connect to that code, and directly driving the writeable variables for the control surface positions. But beyond that we are stuck as far as any additional updates to make the Asobo AP work better.

Edited December 6, 20205 yr by marsman2020

26 minutes ago, marsman2020 said:

You can't fix the current broken heading Asobo AP behavior of "rolling to wings level starting 30 degrees away from the target" with any accessible variable in the .cfg files or SimConnect variable or variable accessible via C++.  The Asobo AP SimVars accessible via SimConnect or C++ are all read-only values, and the aircraft .cfg settings for PID don't fix the behavior due to inaccessible changes to the AP internal code which Matt described earlier in this thread.  The updated PID values from Seb fix a separate issue where some aircraft could only bank in one direction.

Of course one could could a complete AP replacement using their own code for the AP, their own cockpit button/glass cockpit code updates to connect to that code, and directly driving the writeable variables for the control surface positions. But beyond that we are stuck as far as any additional updates to make the Asobo AP work better.

True, you cannot stop the internal a/p doing this, but you can over-impose another wings rolling command as that is exactly what we are working on thanks to the lateral thinking and expertise of a very clever chap who is helping me. Early days as yet but it is working in principle. Refining it is another matter, but it looks promising.

26 minutes ago, robert young said:

True, you cannot stop the internal a/p doing this, but you can over-impose another wings rolling command as that is exactly what we are working on thanks to the lateral thinking and expertise of a very clever chap who is helping me. Early days as yet but it is working in principle. Refining it is another matter, but it looks promising.

I'm looking forward to this!