Progressive pitch oscillations in cruise with AP on

[alexcolka]

Hi,

Sorry if this is covered someplace else, kind of new arround. Ok, so when I'm in cruise with AP on, say at FL250 and 300 kts (my average FL flights) and ALT on, sometimes I get horrid and progressive pitch oscillations becoming from soft to really strong. Also most of the times this happens to me when I reach any given waypoint and turn to the next, then the pitch oscillation starts. By the way it happens to me in both XP9 and XP10, any help is appreciated.

Thanks,

Alex

[philipp]

Hi Alex,

we had several of those reports during the early beta-phase of X-Plane 10, and some time in 10.03 or 10.04 this was fixed (it occured with other planes as well). However, most users had to delete their preferences to reset all settings, which helped anyway with a lot of problem during the beta-phase. So I suggest you make sure you are using X-Plane 10.04 final and start with a fresh set of preferences. This has worked for 100% of the reported AP problems so far.

Philipp

[Japo32]

Yes.. please remove the prefecences folder inside output folder fo x-plane.

Thanks!

[alexcolka]

Thanks Xavier and Philipp, I removed preferences in XP10 (10.04r3) and still no luck, but I'm still testing. Is this applicable to XP9 (9.70) as well? In XP9 this problem is less noticeable but happens from time to time. That's why I prefer XP9 until XP10 gets mature.

Thanks and regards,

Alex

[Japo32]

increase the solutions per frame. Normally it is in 1. Change it to 2.

It is where you choose to have engines running or not when loading a plane.

[PWJT8D]

This happens to me when I get within twenty knots of the red line while in cruise. It seems as when the speed gets closer to the overspeed indicator, this results in more lift then the autopilot can correct for. The aircraft pitches up and it bleeds off the speed, but then pitches down to get back to the assigned altitude. It then becomes an oscillating cycle until the engines are fully retarded until the aircraft can settle down. Then I usually set the throttle limit back farther then I would wish, but it keeps it under control.

[Japo32]

Yes.. that is normal. The pitch up when closing to the max speed is because to not overspeed and break the plane. Then once it goes up.. the speed goes down so it tries to catch the altitude again.. so gain speed.. so it will pitch up automatically to gain reduce speed.

What it happens to you it is due a bad pilot manouver. You don't have to retard the speed when cruising. always 15 knots from red line is good. 20 better.

[Stefan]

I can confirm, that this problem in XP10 occurs only within 20knots below the overspeed indicator - at framerates above 30 it seems to be idependent from the resolution-rate. but it's still not clear for me, why. i cannot see any aerodynamic reason for this extremly non-linear behaviour.

the oscillations also occur, when you deploy flaps, when established on an ils-glideslope, and the current speed shifts into the 20kn-band below the new (reduced) speed-restriction, whis may result in missing the approach.

an intersesting fact ist, that it is very easy to stop the oscillations manually after disconnecting the AP - imho this shows that the AP is involved in some way in producing the oscillations. when you look, what happens at the pitch-trim, during the oscillations, you can see, that the trim still pitches up more, when, the airplane already rises the nose - which is an indication for incorrect controller-parameters (not enough D-component in an PID controller).

But, the most important question is: how does the real CRJ behave in this situation - is the simulation realistic?

Edited March 23, 2012 by Stefan

[Japo32]

When you deploy the flaps.. the maximun tolerance speed red bars comes down.. so if you are at a very speed.. limit to the maximum you will have overspeed and the plane will climb to reduce speed.

These all issues you describe are because bad pilot manouvers. You have to maintain good speeds and not try to reach the maximun speeds, or near the maximum speeds.

If you do, you won't have any problem.

Of course there will be people that will have same issues but that doesn't mean is a bug. Others are flying the plane without any incidences here, because they fly the plane taking care of the speed they are having. That is why I chose to have the throttles manually and not driven by any autopilot servo.

[Stefan]

Hi Japo,

first the bad news:

i am sure that there is a problem with the AP (i do not fly bad manouvers, and i always try to maintain speed as exactly as possible) - i did a lot of systematical testing to proof this.

when you try the following first scenario:

speed 330kn (10kn below maximum), vertical mode ALTS at 9000ft, hold HDG,

then you initiate an 30 degree turn with the AP, the AP loses control over the airplane, when it should stabilize pitch after rolling out from the turn.

second scenario:

speed 330kn (10kn below maximum), vertical mode VS 0.1(very slow climb!) for climbing from 9000ft to 10000ft, hold HDG,

then you initiate an 30 degree turn with the AP and in this scenario the AP has full control over the airplane, no oscillations - maybe it's not able to hold the climb-rate correctly which does not matter.

so, why does the oscillation-problem only occur in the ALTS and ALT CAP mode and not in the VS-Mode?

because they use different controllers! and the controller for the ALTS-mode doesn't work correctly, but the controller for the VS does.

i used 330kn for better testing, but the problem also occurs at 300kn in some situations.

now the good news:

i found a solution (i like finding solutions much more than complaining about problems :-), which works pretty good up to 330kn.

i changed the parameters for the pitch-controller: special/set autopilot constants/use custom autopilot constants

pitch prediction: 2,4 seconds

pitch error for full elevator: 28

EDIT: pitch response time: 3 seconds

pitch rate: 5 degrees/second.

since i found this values using my experience with configuring controllers by trying a little bit, and not using scientific methods, i am sure, there exists a better solution (which has to be computed from the aerodynamical data of the airplane), but this first approach works fine.

i have tested this settings in different situations and could not find any side-effects.

Edited March 29, 2012 by Stefan

[Japo32]

Repeat: the problem is 10knots below maximum. You should be 30 below maximum.

[Stefan]

I have mentioned this, that i have tested at 330kn for more significant results,. it still occurs at 300kn, which is an usual operating speed for cruising and becomes worse unter heavy wind conditions.

and as alex writes, there is a difference between xp9 and xp10 - he says "In XP9 this problem is less noticeable but happens from time to time". the problem was very significant in the beta-releases of xp10, but it still occurs on the release version of xp10.

[Japo32]

Then that is something it should be reported to Laminar BUT NOT ON THE CRJ200 to make them change it and solved it.. you have to test it over the 777 default.

Removing preferences and raising the solutions per frame to number 2 it helps.

Edited March 26, 2012 by Japo32

[Stefan]

removing preferences and setting up solutions did not help.

i have tested now the 777 - it's absolutely stable up to vmax -10kn (which are 380kn) and starts to limb a little bit at vmax-5kn.

so it cannot be an issue of xplane10, because the AP-parameters are airplane-specific.

so i took the autopilot parameters from the 777 which are 17 for the pitch error, 2,0 for the pitch rate 2,6 for the response-time and 2.4 for the prediction time, and used them for the CRJ200, which made the CRJ AP now stable up to 330kn - no more oscillations! maybe crj-development should think about changing the default parameters of the plane for the next release.

[Japo32]

The 777 has not any custom autopilot. If I go to artificail stability and go to autopilot tab.. there "use custom autopilot constant" is unmark. So I don't know where did you pick up those numbers.. (if i click over custom.. others numbers appears)

But anyway.. I have heared about some user having problems that usually are because their bad calibration of Joystick (that is the because of deleting the preferences). I cannot put the same parameters of the autopilot of a 777 to a CRJ200 for sure!

If it goes well for you.. then good.. but that woundn't be a CRJ200. The CRJ200 is called "the Rocket" because it's enourmeous pitch angles it can reach in climbs, and also is a very nervous plane.

I preffer not to touch anything because unless your case, in other pitch oscilations we always found the reason and wasn't the CRJ200. Even I never touch this since the very first 1.0 version and almost no-one but when xplane10 came had thing problem.... and not a few of people.

Edited March 27, 2012 by Japo32

[Stefan]

yes, i know the crj200 as a very special plane. i know it very well from many business and holiday-trips, and i like it. that's the reason, why it want to get working very well on my sim. i agree, that's not an crj anymore, when you change the parameters to much, but it's also not an crj, when it' bucking like a wild horse after a turn with 300kn :-)

btw. there are in fact very small oscillations on the roll-autopilot in the real crj .you cannot feel them, but when yout watch the ailerons, you can see them permanently oscillating a little bit - i could not see this at a real boeing or airbus.

i also thought about the joystick-calibration as a possible source for the problems, because there a very different systems.

it makes a big difference wether the joystick centers by spring-force or provides a "floating" center by the feedback force.

the second difference is, that some trim-systems shift the center of the joystick, others not - i think the crj does not (at least at my joystick).

i have attached a screenshot of my 777-ap-page just for information how it looks like on my system

so for the moment as workaround i will try t find a set of parameters, which provide a stable flight with minimum changes of the default in my environment - this is ok for me.

[alexcolka]

i changed the parameters for the pitch-controller: special/set autopilot constants/use custom autopilot constants

pitch prediction: 2,4 seconds

pitch error for full elevator: 28

pitch response time: 6 seconds

pitch rate: 5 degrees/second.

Hello,

I tested the parameters changed by Stefan and in fact worked! the pitch oscillations in cruise with AP on ALT on are almost gone or hardly noticealble so I'm very pleased with the results and I can now enjoy this great aircraft in XP10 as I can do it in XP9! Great work Stefan!

Many thanks to Javier, Philipp and Stefan for all your help!

Alex

[philipp]

We will need to test this with our standard test flight to see if it doesn't cause any side-effects for us. But if it is fine, we can certainly include it in the future.

Philipp

[Japo32]

Ok.. could you please both of you do the testings in your flights and try to find if everything is ok? Sadly I cannot dedicate long flights to test it... But it is desirable to have a more nervous plane than a 777.

For sure something changed in xplane10 that for some users are making this issues. With a few the joystick good calibration fixes the problem.

I believe it is a problem of frames - performance. All the flight dynamics are calculated per frame. If you don't have a good computer then you will have maybe 20 frames that maybe is enough for your eyes.. but not for the flight dinamics.. In XP9 all was faster.. now in XP10 we have less FPS for now. That is why it suddendly finds this "steps" in flight calculation and "fall" into the pitch oscilations. And that would explain that not all users are having this issue. Those with good computers don't have that. I have good computer and I don't find these issues.. so cannot test it because in all my flights all was perfect.

Changing the Autopilot to become more stable.. like the 777 you make "bridges" in those "holes" the flight calculation finds when having not good performance.. BUT I wouldn't like to see a CRJ like a 747.

Well for now is a good hotfix for the problem.

PD: PIN the topic to follow all testings.

Edited March 27, 2012 by Japo32

[alexcolka]

Ok Javier, for my side definetly I WILL continue testing this because I like a lot this plane in XP10, by the way my joystick settings are optimal (I'm a very meticulously person) and I'm getting arround 30+ fps with eye candy settings in XP10 with the CRJ with 2 frames in flight dynamics settings.

Thanks,

[Stefan]

Javier, i can understand your doubts regarding to changes of the default settings of the CRJ - but the changes do NOT affect the flight-dynamics - which i think are very good (as far as i can estimate this) - but only affect the AP-controller, which should perfectly match the flight-dynamics of a specific plane.

i will do some tests with reduced changes of the default-settings, because i think the parameter which is mainly involved in the problem is the pitch-error-parameter. my modifications result in a faster response of the elevator, which would be very reasonable for this "nervous" plane - as you call it.

the oscillations result from an controller, which is responding to slowly and so always overshoots the correct elevator-trim position. the controller-response cannot be made faster by increasing the res/frame, because it's based on real-time. (but the problems, which occured in the 10.x.beta-releases COULD be fixed by higher res/frame - but they seem to be fixed now in the release-version)

unfortunately the solutions per frame have no influence on the oscillation-problem in my configuration. at the moment i use 3 res/frame@60(!)fps, which means 180 resolutions/sec, i think. but as you say, there may be a drop of the framerate from xp9 to xp10, which may affect the autopilot in some way. or may be the evaluation of the flight-dynamics have changed in some way, making the plane respond faster.

but i have no xp9 - and no time - at the moment to do investigations on this.

[Japo32]

I have included the acf file modified with autopilot in user modification folder for next update. I don't want to include in the default crj200.acf file because you don't know how many times we changed things that supposed to be totally "soft" for the plane.. and made big bugs after a while.

Not all people is having this Pitch issue so I will leave that file for those poeple that have the problem.

I put the 4 pitch parameters you told me in that file.. but for me the pitch responce time be changed from 1.5 to 6 seconds is too much. I believe a CRJ200 would be 3 seconds or so.

[Stefan]

yes, 6 sec is to much - the 777 has 2.6

[Japo32]

You told 6 in the numbers you tried.

[Stefan]

yes, 6 works, but this was the first try, next try were the parameters of the 777, which result in smaller changes of the default crj-settings and also provide good results. i want to keep the changes as small as possible, but have not tested all combinations of settings so far, because it needs a lot of time. it seems the prediction-time and the pitch-error are the most important parameters here, and maybe the outhers need not be changed - but i'm still testing...