Board index FlightGear Development Aircraft Autopilot and route manager

747-400 A/P Help

Designing a stable autopilot is one of the hardest things. Need help?

747-400 A/P Help

Postby swwon » Mon Jul 25, 2016 3:08 am

Can someone help point me in the right direction to fixing the autopilot on the 747-400? I am primarily looking to test with FG 2016.2.1, since I've heard someone is working on updating the "400" already, but I can't find anything on its progress. Is anyone doing anything with it currently? Anyways, Is there a way to copy or port over the A/P from the 777 or maybe even the A350 system? Any assistance would be appreciated.
I've read the autopilot "How to" Wikis but because of my lack of programming knowledge I didn't get it. I was hoping it would be easier to port over (copy/paste) the lines from Nasal and XML from an other properly working JSBsim A/P would do it, but I'm obviously overlooking somethings because it doesn't seem to work. Can someone give me a hint at where I need to start? Thanks
After that I'd like to start cleaning up the cockpit.
But first things first, please help me fix my 747-400.
swwon
 
Posts: 232
Joined: Sun Nov 22, 2015 10:32 am
Version: 2016.2.1
OS: Windows 10

Re: 747-400 A/P Help

Postby Octal450 » Tue Jul 26, 2016 2:56 am

APs aren't classified as JSB or Ya, my IT-AUTOFLIGHT flies both Yasim and JSBs fine. (MD-88-Yasim, KC-137R-JSB)

It's very simple to install, if you want to try it out. Note that's it's still in beta, V3 is still in progress. I can give you instructions if you want.

The A350 AP runs the V2 version of IT-AUTOFLIGHT. I will update it to V3 when it's more stable.

Josh
Waste of time. Goodbye forever.
Octal450
 
Posts: 4398
Joined: Tue Oct 06, 2015 12:51 pm

Re: 747-400 A/P Help

Postby Thorsten » Tue Jul 26, 2016 5:50 am

APs aren't classified as JSB or Ya, my IT-AUTOFLIGHT flies both Yasim and JSBs fine. (MD-88-Yasim, KC-137R-JSB)


There's different technologies around. JSBSim allows you to write a native AP, FG has generic tags for the same purpose which would also work in YaSim. Conceptually if you deal with a JSBSim plane, writing the AP in JSBSim can be desirable because then you can run and test everything in JSBSim standalone, which saves you lots of time over having to use FG as a container all the time.

Conceptually there's low-level tasks (damping oscillations, keeping wings to a commanded bank angle) as well as high-level tasks (aim the plane to follow a series of radio beacons).

In general you can't expect that the low-level tasks port well - the gains required to deal with oscillations may be rather different for an airliner and a single engine prop. If a plane is naturally stable, the AP never needs to manage stability, if not then the story is quite different.

The high-level tasks port better, but also the turn radius you need to anticipate when changing to a new leg on a route depends on the flight characteristics, climbing rate does,...

So while the generic structure of what the AP does is usually similar for similar planes, copying a framework rarely seems to lead to really good results and a good AP has to be custom-tuned to the specific plane.

Mathematically an AP is a fairly complicated set of coupled integro-differential equations, so you can't expect a simple solution really.
Thorsten
 
Posts: 11123
Joined: Mon Nov 02, 2009 8:33 am

Re: 747-400 A/P Help

Postby swwon » Tue Jul 26, 2016 7:29 am

Thanks Thorsten, understanding the complications, maybe I'm out of my depth here. I was hoping there was a simpler port over and just adjust the calculations based on the "bigger airplane's" characteristics. I'm a huge fan of the 747-400.
swwon
 
Posts: 232
Joined: Sun Nov 22, 2015 10:32 am
Version: 2016.2.1
OS: Windows 10

Re: 747-400 A/P Help

Postby swwon » Tue Jul 26, 2016 7:32 am

it0uchpods wrote in Tue Jul 26, 2016 2:56 am:APs aren't classified as JSB or Ya, my IT-AUTOFLIGHT flies both Yasim and JSBs fine. (MD-88-Yasim, KC-137R-JSB)

It's very simple to install, if you want to try it out. Note that's it's still in beta, V3 is still in progress. I can give you instructions if you want.

The A350 AP runs the V2 version of IT-AUTOFLIGHT. I will update it to V3 when it's more stable.

Josh

Hey well if you could PM me instructions or post here that would be great. Is there any other program I need other than Notepad++? I'm fine with V2 for now.
swwon
 
Posts: 232
Joined: Sun Nov 22, 2015 10:32 am
Version: 2016.2.1
OS: Windows 10

Re: 747-400 A/P Help

Postby Octal450 » Tue Jul 26, 2016 3:50 pm

@Thorsten
Hmm, did not know that. Thanks.

@swonn
I'm trying to finish it up, but between surgery on my feet and other stuff, I'm quite busy. I am writing a full guide the will be finished in V3. I am not sure how it will fly the 747-400, the FDM always was problematic for me to handfly. I installed it out of curiosity, I'm testing, and then I'll push to a fork on my repo.

Josh
Waste of time. Goodbye forever.
Octal450
 
Posts: 4398
Joined: Tue Oct 06, 2015 12:51 pm

Re: 747-400 A/P Help

Postby Octal450 » Tue Jul 26, 2016 4:08 pm

Yeah this FDM is really weird. It does not fly correctly. Due to this, my AP can't fly it properly either. the 747-800 has a nice FDM, that one would probably work better. I don't have the time to tune it to the 747-400 right now. Sorry.

Josh
Waste of time. Goodbye forever.
Octal450
 
Posts: 4398
Joined: Tue Oct 06, 2015 12:51 pm

Re: 747-400 A/P Help

Postby swwon » Tue Jul 26, 2016 5:41 pm

Hmm, interesting, maybe a rebuild of the whole thing is necessary...
Anyways, I'll keep looking for ideas. I wonder why some planes are easier to correct than others?
swwon
 
Posts: 232
Joined: Sun Nov 22, 2015 10:32 am
Version: 2016.2.1
OS: Windows 10

Re: 747-400 A/P Help

Postby Thorsten » Tue Jul 26, 2016 6:11 pm

I wonder why some planes are easier to correct than others?


Because FDMs can be realistic or cartoons.

If you take the FG SPace Program Shuttle , it has an FDM straight out of Aeromatic. It flies nicely, feels a bit like an airliner, and is easy to control 'raw'. You would find writing an AP for it easy.

If you take FGAddon Shuttle, it is based on tons of NASA wind tunnel data and in-flight measurements. It has loss of airfoil effect at hypersonic speeds, mach dependence of basically everything, cross-coupling between control channels and is dynamically yaw unstable. You couldn't even fly this by hand and would find writing an AP for it hard (I think by now it's some 10.000 lines of AP code).

So there's a school of thought in FG to first take great pains to model real aerodynamics (and create something pretty unflyable in the process) and then take more pains to create an FCS which handles it to the degree that it can be used, and there's another school of thought which says what the heck, the user is never going to fly it without FCS anyway, so we might as well dispense with complications and complications to overcome the complications and just do something simple up-front that roughly agrees with published characteristics.

You'll find that the planes with the second approach are easy to control, correct and tune, whereas the planes with the first approach never are.
Thorsten
 
Posts: 11123
Joined: Mon Nov 02, 2009 8:33 am

Re: 747-400 A/P Help

Postby Octal450 » Wed Jul 27, 2016 2:40 am

Yup. Well said @Thorsten

My AP works on a wide variety of planes, but the 747-400 FDM just flies too badly to do that. Full trim down, and the thing still at 15deg up!

:(
Waste of time. Goodbye forever.
Octal450
 
Posts: 4398
Joined: Tue Oct 06, 2015 12:51 pm

Re: 747-400 A/P Help

Postby Thorsten » Wed Jul 27, 2016 6:12 am

My AP works on a wide variety of planes, but the 747-400 FDM just flies too badly to do that.


May be a stupid question, but... weight distribution? I remember the Concorde isn't really trimmed aerodynamically at all but by pumping fuel back and forth to shift the CoG.
Thorsten
 
Posts: 11123
Joined: Mon Nov 02, 2009 8:33 am

Re: 747-400 A/P Help

Postby swwon » Wed Jul 27, 2016 9:23 am

More very interesting stuff. I personally believe the 747-400 actually flies pretty realistically hand flying it, it's weight it's transfer of angle of attack etc.. The 747-8 in comparison seems too "easy" and nimble for a plane of its size.
swwon
 
Posts: 232
Joined: Sun Nov 22, 2015 10:32 am
Version: 2016.2.1
OS: Windows 10

Re: 747-400 A/P Help

Postby Octal450 » Wed Jul 27, 2016 12:36 pm

@Thorsten
Hmm maybe, there's no "Leveling" or "Balence" thing, or something that says me the COG, so I don't know. (but 2 Fuel and Payload options????)

@swonn
The 747-8 is vertically to nimble. Laterally, it flies pretty accurately from what I can tell -- never flown an 747 or 747sim.

The issue with the 747-400 is that the FDM tries to emulate fly by wire, but I think that is not lovely to do in the FDM!!!!!!!!!!!!!!!!!!!!!!!!!!! (well, maybe you can, I don't know.)

My point is, my AP can't fly is, because when he does, he eventually put's full trim down, and the plane is still at 15+ pitch! So......

In a later version I plan to use the elevator, as per real life, and then have the trim adjust to keep the elevator centered.


My conclusion on the 747-400 is that it is unfinished. If someone decides to work on it, then maybe all this will be solved! (is someone? I don't know) The 747-8i is unfinished, but stable, so These things don't occur.

Josh
Waste of time. Goodbye forever.
Octal450
 
Posts: 4398
Joined: Tue Oct 06, 2015 12:51 pm

Re: 747-400 A/P Help

Postby Thorsten » Wed Jul 27, 2016 1:36 pm

I'm not sure trim is an FDM issue though. Usually aircraft don't have special trim surfaces, the trim either adjusts springs for the elevator zero point, or in a fly by wire setup is a bias to the elevator command.

Either way, it's probably an FCS issue how trim and elevator command are merged and how much authority trim is given over the airfoils. But I'm fairly sure that in reality the end effect of both trim and stick is that the elevator surfaces move. So if the plane flies okay by hand, then there's likely nothing major wrong with the FDM as such.

So you could simply look into the FCS and see how trim and elevator-cmd are merged.
Thorsten
 
Posts: 11123
Joined: Mon Nov 02, 2009 8:33 am

Re: 747-400 A/P Help

Postby AndersG » Wed Jul 27, 2016 2:02 pm

If this is the 747-400 by Gijs it may be that the trim channel actually adjusts the horizontal stabilizer rather than bias the elevator position. I have some faint memory of discussing the pros and cons of such an approach with him on IRC some (~6-7?) years back. I don't recall the 747-400 systems but on older jetliners these two channels (elevator and stabilizer) would be isolated FC channels and if the stabilizer is not appropriately set for the situation you can easily run out of elevator authority.
Callsign: SE-AG
Aircraft (uhm...): Submarine Scout, Zeppelin NT, ZF Navy free balloon, Nordstern, Hindenburg, Short Empire flying-boat, ZNP-K, North Sea class, MTB T21 class, U.S.S. Monitor, MFI-9B, Type UB I submarine, Gokstad ship, Renault FT.
AndersG
 
Posts: 2448
Joined: Wed Nov 29, 2006 9:20 am
Location: Göteborg, Sweden
Callsign: SE-AG
OS: Debian GNU Linux

Next

Return to Autopilot and route manager

Who is online

Users browsing this forum: No registered users and 1 guest