Autopilot parameters

[lonewulf 1]

Hi all. I'm new to this forum and I hope I'll find my questions answered here. I'm desperately trying to improve some of the autopilot (and autothrottle) behaviour on the default FS2004 heavies. AAM shows some relevant parameters in record #1199 but unfortunately changing of these parameters have no effect whatsoever. My primary concerns are the sluggish heading rollout and speed adjustment. Although there ARE some specifically mentioned parameters for heading hold/rollout and for autothrottle none of them shows any effect when I alter them.Where's the secret?:-hmmm Regards Oskar

[n4gix 4,947]

Hi Oskar! Welcome to the forum... Hopefully someone here will pipe up with some tips, seeing as how I steered you here to begin with... ;)

[lonewulf 1]

Thanks Bill! But obviously again nobody seems to know anything about these specific issues..... ;(It's the same strange thing that I observed many times. Everything must be polished-up tp the max but when it comes to flying parameters nobody seems to care. Maybe it's my own fault. I'm seemingly still too much into the FLYING thing rather than the PLAYING thing ;-) Oskar

[Guest iholrf]

>Thanks Bill! But obviously again nobody seems to know>anything about these specific issues..... ;(>It's the same strange thing that I observed many times.>Everything must be polished-up tp the max but when it comes to>flying parameters nobody seems to care. Maybe it's my own>fault. I'm seemingly still too much into the FLYING thing>rather than the PLAYING thing ;-) >OskarI wish I had an answer for you but all I can provide is some moral support with the knowledge that some of us do care, but like you don't know how to solve the problems (yet :)). Sadly, I am discovering there are some problems that may not be solvable in the current version.I got into the guts of the sim to learn about it because there are so many aircraft that are just so wrong it was driving me nuts (even default AC). So I decided to build my own. Like you, I also think to much care is put into shiney surfaces and detailed passenger cabins (the "polish" you describe). There is too much empahsis on things that actually distract from the flying experience and not enough on the accuarcy of the flight model, or usefullness of the panel. The Lear being a prime example... here is a plane that should be joy to fly but is one I quite frankly hate.An aircraft is for flying not looking at.All I can suggest is keep looking, don't give up and share what you learn so more of us can get our projects up to the level of quality you would also like to achieve.Best of luckShad

[Guest Ron Freimuth]

"AAM shows some relevant parameters in record #1199" REC 1199 in the AIR file has been dead with FS2K2 and later. Thus, one is limited in what autopilot control one has in the [autopilot] section of aircraft.cfg. Ron

[LAdamson 240]

>An aircraft is for flying not looking at.>My opinion......The flight dynamics are an illusion, to mimic those of the real aircraft. Yet, I rate exteriors & panels just as highly, to complete the illusion. I'm basing this on my background of real aircraft construction, the Experimental Aircraft Association, and involvement with warbird restoration...............where well crafted airplanes, mean something.Of course this has nothing to do with auto-pilots :D L.Adamson

[lonewulf 1]

Thank you Ron, that's what I was afraid of... x( . Have you any information about kind of PID variables for heading and speed control available in the CFG file? MIcrosoft's SDK doesn't mention any of these.Oskar

[Guest iholrf]

>>An aircraft is for flying not looking at.>>>>My opinion......>>The flight dynamics are an illusion, to mimic those of the>real aircraft. Yet, I rate exteriors & panels just as highly,>to complete the illusion. I'm basing this on my background of>real aircraft construction, the Experimental Aircraft>Association, and involvement with warbird>restoration...............where well crafted airplanes, mean>something.>>Of course this has nothing to do with auto-pilots :D >>L.AdamsonAbsolutely. I completely agree. But (there is always a but eh? ;)) while a well crafted exterior model is impressive (I think SimTECH's Beechcraft Model 18 is stunning), it needs first to A) fly properly and :( make me feel like I am inside an aircraft _in the pilot's seat_. I, as a pilot, spend 99% of my time in, well, the cockpit. I do not spend it looking at the cabin wing view. (I have seen enough of this as an airline passenger, heheh)This is where 99% of the Freeware addons leave me flat and why SimTECH's Model 18 is relegated to the "interesting but not flyable" folder of my AC archive. I should point out I am a huge fan of the Model 18. It is IMO one of the most beautiful AC of all time (the Volpar conversion of the Model 18 is the ugliest AC of all time. Its just so wrong, but I love it anyway and is what I am building). As a budding panel builder, I am somewhat surprised by how bad so many panels are when making one is not actually that hard (campared to the modeling process - at my skill level). I figure most model builders are burned out by the time it comes to making the panel. So I thought I would start from the inside and work my way out.This comes from years of expeience in the Airforce, my own piloting and 35 years of fascination with flying and sailing machines of all kinds (seeing as how were listing credentials heheh :))Shad

[Guest Ron Freimuth]

>Thank you Ron, that's what I was afraid of... x( . Have you>any information about kind of PID variables for heading and>speed control available in the CFG file? MIcrosoft's SDK>doesn't mention any of these.>Oskar DL my 'comments on aircraft.cfg' at:http://perso.wanadoo.fr/hsors/FS_Soft/fsdocs.html I added some more info on the FS9 [autopilot] section just last month. I've set 'throttle rate' as high as 5.0 at times. That still doesn't help SPD hold a lot, but helps a bit if it's still stable. A slow Powerplant spool up time also increases the time it takes for SPD to change and stabilize. Also, 'gs proportional gain' is 9.52 by default. 15.0 or 20.0 helps most any AC a lot as far as holding the GS. Actaully, it's possible to program P, I, D functions even in XLM gauges. I have a test version working. I didn't program it, but gave the guy who did ideas of how to proceed. SPD hold would be a good candididate for simple Feedback Control. It might be good enough to just take "SET IAS" - "FS IAS" and multiply by some Gain constant. Then, use that error signal to set the Throttle. However, some D would allow higher Gain, thus faster and closer control. While I(ntegral) FeedBack would be necessary to eventually reduce the SPD error to zero. Ron

[lonewulf 1]

Thank you RonYour comment is highly appreciated. I've downloaded your 'comments on aircraft.cfg' and I see my own findings confirmed so far. I have looked up to the XML-gauges of the default A/C (B747) and it might well be that there's something like a proportional control in the autopilot speed control section. But unfortunately I'm not familiar with XML programming so it looks rather chinese to me. So it could well be that this simple control is responsible for the veeery sluggish behaviour of the autothrottle. After all it looks like a worst-case scenario to me as I have to learn about XML programming and that will be a nightmare as I see it....;( (My highest level of programming being Visual Basic, but maybe I'll find a decent tutorial which will not overload my old brain...)BTW for better engine response I have changed record# 1505 to a more linear function with impressive results.As I would figure with the much faster throttle response a P-I control could work quite fine whereas the D control would only be necessary if oscillations occur.Oskar

[LAdamson 240]

>>This comes from years of expeience in the Airforce, my own>piloting and 35 years of fascination with flying and sailing>machines of all kinds (seeing as how were listing credentials>heheh :))>I should have added that I flew R/C "tower view" for many years also. That means I sim fly at least 50% from the "outside"... :D L.Adamson

[Guest Ron Freimuth]

>Thank you Ron>Your comment is highly appreciated. I've downloaded your>'comments on aircraft.cfg' and I see my own findings confirmed>so far. I think I got some comments from someone else on setting the 'derivative limit' to 45 to 90 degrees in that aircraft.cfg file -- so it is active when one isn't close to the VOR radial. If you discover useful things about the FS9 [autopilot] lines let me know. > I have looked up to the XML-gauges of the default A/C> (B747) and it might well be that there's something like a>proportional control in the autopilot speed control section. The real autopilot is in sim1.dll. The panel gauges only control the inputs to it. However, one or more of the 'instruction' files have some adventure programming language code that does speed control independent of the FS autopilot.>But unfortunately I'm not familiar with XML programming so it>looks rather chinese to me............> (My highest level of programming being Visual Basic, but maybe>I'll find a decent tutorial which will not overload my old>brain...) At least I was familiar with the XML 'RPN' from using an HP RPN calculator. Much of the FS RPN is virtually identical to the HP calc. That still doesn't make RPN easy for me to program. I like it for calculator input, but not for program code. Regardless, plenty of people in this Forum seem to have mastered it.>BTW for better engine response I have changed record# 1505 to>a more linear function with impressive results. Yes, REC 1505 has the effect of controlling spool time. Though that is a side effect of it's effect on 'Corrected Fuel Flow'.>As I would figure with the much faster throttle response a P-I>control could work quite fine whereas the D control would only>be necessary if oscillations occur.>Oskar REC 1199, now defunct, had SPD hold 'gain', 'derivative', etc. In FS2K I could considerably improve SPD hold. Getting to set SPD much faster than the default settings. Leave it to the MS guys to delete such desirable AIR file records -- and just when we were figuring out how to set them. BTW, MS messed up the SPD hold in FS2K2, especially for IAS hold. They apparently use GS rather than AIR speed (IAS or Mach) for the Derivative FB. This is especially bad when there are winds aloft: the AC speed will increase or drop untill the GS Rate of change is low so it no longer messes up the D feedback. Nor did they fix it in FS9. I have to wonder how much these MS guys really fly FS AC. They should have noted the problem before FS9 was released. A few people in the FS Forum have noted the problem. Ron

[lonewulf 1]

Hi Ron> If you discover useful things about the FS9 [autopilot]>lines let me know. You bet I will...... However as I see it now there's obviously no chance to work around an improved speed/heading capturing which is - see my note at the end of this text - simply ridiculous.> However, one or more of the 'instruction' files have some>adventure programming language code that does speed control>independent of the FS autopilot.Are you referring to the *.ABL files? They look very much like PASCAL source code. However I wonder about that as PASCAL is not an interpretable language.> BTW, MS messed up the SPD hold in FS2K2, especially for IAS>hold. They apparently use GS rather than AIR speed (IAS or>Mach) for the Derivative FB. This is especially bad when>there are winds aloft: the AC speed will increase or drop>untill the GS Rate of change is low so it no longer messes up>the D feedback.>> Nor did they fix it in FS9. I have to wonder how much>these MS guys really fly FS AC. They should have noted the>problem before FS9 was released. A few people in the FS Forum>have noted the problem.>> RonYoure exactly right. I think we have to deal with a main problem here. As my background is purely flying (I'm a retired airline captain) I am of course more concerned about oddities in respect of dynamic behaviour of the flight model rather than of scenery issues or shiny surfaces..:) IMHO the FS developer team has a few shortcomings which are not easy to define: Either they don't have competent beta-testers or they simply ignore their inputs. Maybe also the timelines are too dense for significant changes after beta-testing. There are a few real "stupid" errors/misbehaviours in FS9 (also in earlier versions) with regard to the "real" flying world which - given the many parameters involved in the flying model - could easily be corrected. So the simple question remains: Are they not aware of the most simple issues of flying? Arey they just ignoring any inputs to detected errors/misbehaviour? Or - most likely - do they simply not know about some weak spots in FS?:-hmmm Oskar

[Guest iholrf]

> There are a few real "stupid">errors/misbehaviours in FS9 (also in earlier versions) with>regard to the "real" flying world which - given the many>parameters involved in the flying model - could easily be>corrected. So the simple question remains: Are they not aware>of the most simple issues of flying? Arey they just ignoring>any inputs to detected errors/misbehaviour? Or - most likely ->do they simply not know about some weak spots in FS?:-hmmm >>OskarI think you may be a bit optopmistic about how easy things are to correct. I have been following the development of X-Plane and Flight gear, and all I can say is at least the autopilot works in FS9, if not perfectly. In X-Plane there are a lot of autopilot issues (normally flakey), and in FlightGear, there was none (as of the last release I looked at).I don't know if you followed the development of FS9, but the team was not huge, and MS has never in the history of its company realeased a polished, complete and bug free product (nor on time). Not one. It is one of the reasons they are so successful (even though on first glance it may seem to be contrary to logic). They work on a philosophy of "good enough, now get it to market". Their products are howerver always pretty and packed with useless features which sells their product to the uneducated, and for the most part, uncaring consumer public (I don't mean that as an insult to my fellow humans, it is just reality). It seems like MS follows the GM marketing philosophy of the 30's: give them a cheaper car packed with more toys for the same money as their competitor (GM soon outsold Ford, even though in many respect their product was inferior).In the end though, FS9 is a game, not a Sim, and it makes no claims to the contrary. It models many things far too incorrectly to be considered a true simulator though it is improving. (Even a brick will fly in MSFS).My opinion only, and I do understand your frustration (it did take them a year to fix the huge, huge, huge memory leak that should not have been present to start with).Shad

[n4gix 4,947]

>Youre exactly right. I think we have to deal with a main>problem here. As my background is purely flying (I'm a retired>airline captain) I am of course more concerned about oddities>in respect of dynamic behaviour of the flight model rather>than of scenery issues or shiny surfaces..:)See Oskar! I told you that you'd at least get a good conversation started here...You're being too modest though, being such a high-hour Airbus Computer Operator, oops!... Pilot, and all... ;)