Issues with Autopilot

[Ryan Mann 0]

I am having an issue that I've possibly tied to AAO w/ pmdg 737. After takeoff, if I try to engage autopilot, it does but doesn't quite listen. Altitude assignments never work. Heading doesn't listen to bank angle and slowly turns no matter what and the autothrottle is all over the place. I'm not sure where to start as I'm pretty sure I've read the manual correctly on how to set up. It responds to my input and mapped buttons. I originally thought it to be a pmdg issue and saw others recently having issue w/ the AP. But I disconnected from simulator, went into controller settings in MSFS and switched back to default for my airbus joystick and tested and it listens and follows my autopilot instructions. I understand this could be tricky but I'm hoping someone could point me in the correct direction on where to start. I was also using the throttle quadrant. I also have been attempting to get the streamdeck plugins to work on my xl from lorby as well as chris_a320_pu plugins from flightsim.to but have put that to the side while I tried to figure this out. It is a clean install of the OS, clean install of MSFS, only using pmdg 737 and AAO at the moment. nothing else has been installed yet. thank you

 

Ryan Mann

[Lorby_SI 2,042]

Sounds weird. I can't imagine that it could make a difference for the autopilot by what method a button has been pushed. I wouldn't know how to mess with the internals of the AP even if I wanted to, that seems pretty much impossible.
Did you configure your controls using the ROTOR_BRAKE pattern or the native PMDG SDK events? Trouble with the ROTOR_BRAKE scheme is, that most implementations I have seen ony send the "mouse down" event, but don't send another to release the virtual mouse button again. In the long run, that could lead to problems - but not in the way that you describe. For the record, AAO and the PMDG DC-6  or 737 work perfectly fine for me.

Word not allowed, I dimly remember a similar question months ago. I think that it had been resolved too, but I don't remember what the user did wrong or what he did to mend it. Or even where that question was asked. 

Gut feeling says to check for any anomalies with the trim or with any axis assignments in general. If "something" would constantly override control surface or throttle inputs, I suppose the AP might be in trouble.

Edited January 13, 2023 by Lorby_SI

[Ryan Mann 0]

I should've elaborated more, I apologize. I was manually clicking on the autopilot, heading, speed, FDs, AT, etc. The only actions that were bound to my airbus quadrant/joystick for autopilot were autopilot toggle, lnav, vnav, but I was not actually using them for this issue. I would take off and manually set these things with mouse clicks and they just didn't listen although they did "activate" and turn green for everything. As soon as I disconnected AAO completely and manually switch back to the default settings for the joystick, the autopilot acted properly. I guess what i'll have to do is maybe just add one assignment at a time until I figure it out and see which event is the cause. Is there a way to disable each button/axis assignment instead of deleting them to try different scenarios? 

I do have brakes set but nothing advanced or complicated (not sure if it is rotor_brake variable) but I will check. I think it was a different event I assigned and it seemed to work fine. I watched your YT video from few years ago and assigned things like that as well as continuing to read from the manual. 

The trim setting I've left alone b/c I'm just doing a freeflight in this testing, not using the FMC, therefore not setting up the actual trim, so whatever it is when the plane loads on the runway, I don't touch it. Maybe set at 0 by default.  Appreciate the info and I'll see what I can come up with this weekend. I generally fly as realistic as possible and use the FMC, but I just wanna test the app first.

The main reason I wanted to use AAO, and correct me if I'm wrong, was to capture the surface events when replaying or recording video. Any software that I've used kept saying it captured these events but every time I replayed, my gear never went up, flaps never went up, etc. etc. Also I recall someone asking why one would use your software, and you replied back with many different reasons, also one being voice recognition, which interested me as well but haven't even looked into that yet. One thing at a time. 

Thanks again!

[Lorby_SI 2,042]

28 minutes ago, Ryan Mann said:

and they just didn't listen although they did "activate" and turn green for everything.

This sounds just so weird. I cannot imagine how that would even be possible, sorry. I believe you that you are seeing this, but I have no idea at all about a cause. If someone asked me off the bat, I would sternly say that it is entirely impossible... But you never stop learning... Plus, nobody else has reported this, and I have to assume that a lot of people are using this combination. 

What about the other aircraft? Any Global or Aicraft Automated Scripts running in AAO for this plane? What if you switch to a different livery and clean out all the axis and buttons? Did you set everything up manually or did you import/apply 3rd party templates or scripts? Any possibility of interference by MSFS settings, controller assignments or other addons?

Edited January 13, 2023 by Lorby_SI

[Ryan Mann 0]

I did import a script for my SDXL from:

https://flightsim.to/file/34382/pmdg-b737-700-streamdeck-xl-profile-need-lorby-s-axis-and-ohs

I haven't used AAO w/ other aircraft yet. If I disconnect from AAO, does that "shutdown" any scripts imported through it? I will be trying everything from scratch today. I manually set up the axis and button assignments and the only script that I imported (from above link) was for that plugin on the streamdeck. I set up a new profile for the tca airbus joystick and throttle quadrant in MSFS Controller Settings and removed all assignments that matched up with any I was assigning in AAO (in real-time ie. as I was assigning each one, I immediately went into to delete the one in MSFS).

After the issues, I disconnected from AAO only, went into MSFS controller settings, chose the default profile (instead of my new profile) with all the original assignments. Then I took off and tested the autopilot and everything worked again as was expected.

[Ryan Mann 0]

I deleted all the button assignments, issue still happened. Targeted the axis assignments and got it working much better after I deleted my axis assignment for elevator and set that instead in MSFS Controller Settings for Elevator Axis. It was previously set under simulator axis event > elevator in AAO. 

On a side note: my MSFS "crashes" (gives warning on loading that something crashed) on loading. I only have pmdg 737 and aao installed. I temporarily moved 2 folders from community:

lorbysi-content-fippanel
lorbysi-content-hooks

and it no longer crashes. not sure what they are specific too or why they are crashing. 

[Lorby_SI 2,042]

44 minutes ago, Ryan Mann said:

On a side note: my MSFS "crashes" (gives warning on loading that something crashed) on loading.

I've never seen that. What does that look like? Both modules are pretty much essential to the AAO functionality. The "hooks" more than the "fippanel". Remove the "hooks", and I am pretty sure that the StreamDeck profile will stop working.

Look, I'm sorry that you are having these problems, but at the moment you seem to be the only one. There are a couple of thousand people out there using AAO, and I haven't had any complaints regarding the PMDG (except the one I mentioned earlier). A few testers (and myself) are using this combo (AAO and the PMDG aircraft) with zero issues. There must be something else in play that creates that kind of interference - but I have no idea whatsoever what it could be. 
Are you using the PMDG SDK with the autoscripts in AAO? Are your controllers paricularly noisy, so they could be pushing the axis around all the time (you mentioned the Elevator)?

The product description in the shops says in (I hope) no uncertain terms that you have to try the demo version before purchase, to make sure that everything works. Did you do that and did it work then? If it did, what changed?

 

Edited January 14, 2023 by Lorby_SI

[Lorby_SI 2,042]

2 hours ago, Ryan Mann said:

If I disconnect from AAO, does that "shutdown" any scripts imported through it? I

The imported scripts are a feature of AAO, they have nothing to do with the simulator. When the app has been disconnected from the sim, there is no interaction anymore.

[Lorby_SI 2,042]

43 minutes ago, Ryan Mann said:

or why they are crashing. 

Neither do I, never heard of it.

But I had a customer recently, where one of them stopped working out of the blue (which in computer terms is impossible, there is always a reason).

Deleting the folders where MSFS compiles these modules cured that problem:

C:\Users\...\AppData\Local\Packages\Microsoft.FlightSimulator_8wekyb3d8bbwe\LocalState\packages

Edited January 14, 2023 by Lorby_SI

[Lorby_SI 2,042]

I did a short test.

1. Updated to the latest version of the PMDG 737 700.
2. Started AAO, started MSFS
3. I have a full profile in AAO, every axis has been assigned, my SaitekPanels are configured and I have a couple of buttons in there too (trim etc.)
4. I wasn't using the StreamDeck for this test (=the Elgato software wasn't even running, I've disabled it)
5. Put the plane at EDDM 26L, "Fly now", waited for it to compile (took about 10 minutes)
6. programmed the FMC, laid in a short flight plan from EDDM to LIRF, set an initial altitude of 10.000ft on the MCP
7. Took off, engaged Autopilot, LNAV and VNAV
8. The plane followed my selected departure without problems, the Autothrottle worked as expected and the AP captured the 10.000 feet mark perfectly.

My devel computer is running Windows 11, and I have the MSFS beta.

 

Edited January 14, 2023 by Lorby_SI

[Ryan Mann 0]

I appreciate all the time you have taken to help. I did not try the demo, I just purchased it. 

I'd say the rudder is a little bit "noisy"? I look at the data when it pulls to the left on takeoff and I see it jump a lil here and there, but to be fair, I've always had the pulling left issue even with an older joystick, which is the reason I purchased the new one. 

I will continue to test until I figure this puzzle out because I am very much interested in getting it to work. I am very thankful for these types of software and the amazing work and incredible amount of time that developers put into this hobby to make my pursuit of realism worth it in the end. 

Being that the common denominator is me and my setup b/c I do indeed know it's working great for everyone else, it will just take some patience on my end. I appreciate your response and will continue to support. 🙂

[Lorby_SI 2,042]

For noisy input you could try creating a deadzone, dial up the "Filter" value or apply a negative Curve (I have to to the latter all the time, because I find the rudder response way too sudden in MSFS when actuated via SimConnect)

But that is just stumbling around in the dark - assuming that the PMDG might be thrown off by constant controller input, and evening that out would solve the problem. The easiest way to test would be to simply not assign any flight control axis in AAO...

Are you using the PMDG SDK with the associated autoscript running in AAO? 

Edited April 6, 2023 by Lorby_SI

[Ryan Mann 0]

I'll try the deadzone tip.

I am not using the PMDG SDK either. 

[HisBoyElroy 1]

I've had similar issues to Ryan since late last year with the PMDG DC-6, 737-600, and 737-700. In my case, the DC-6 would yaw slowly from side to side when under gyropilot, maybe 4-5 degrees either side of target, in either heading or nav modes. The 737s were unstable in throttle and pitch control under all autopilot modes, constantly pitching up and down and accelerating/slowing past the target. 

Finally made some time to try to sort it out and during the process of elimination, I discovered that closing AAO resolved the worst of the behaviors. I use a VIRPIL flight stick and rudder pedals, which are quite sensitive. I applied a 2% deadband to the aileron, elevator, and rudder axes in AAO, the aircraft are stable once again (subject to normal atmospherics). 

[Lorby_SI 2,042]

Maybe this behavior also depends on how exactly one assigns the flight controls - via variables (selection at the top) or events (second box from the top).

Normally I would expect the internal AP logic to have a "sensitivity" setting, so it doesn't react to the input. But maybe there is a way to bypass that, so the AP is fighting againt the actual yoke input. When in doubt, I would recommend to always use the Events rather than the Variables. Events seem to be the MSFS default most of the time.

Edited April 6, 2023 by Lorby_SI