[PART, 1.0.2] Anatid Robotics / MuMech - MechJeb - Autopilot - Historical thread

[AgentMOO]

20 post in 6 hours by 4 posters ? You re kidding ?!?

AgentMOO : try the build from yesterday. it fix a problem for node with manually activated engines (which is your case in the video you posted)

snip

I just tried it, but still the same result. One less thing it could be =P.

Edit: Just did a clean install of KSP with MJ Dev build 462, created a new sandbox game and moved my craft file in. Last time I did this it worked for only the first launch. This time it didn't even work on the first launch.

But here is a brand new output file: http://www./view/t9u1f75cm7gi1n0/output_log.txt

Sarbian, I really really appreciate all the hard work you put into this. Your a KSP HERO!

Edited May 19, 2015 by AgentMOO

[linuxgurugamer]

Install the latest dev build.

Was this your first launch using ascent guidance with that craft? Remember you have to launch once to the target altitude (fully, circularize and everything) then revert and then launch to rendezvous. Otherwise MJ has no idea when to launch.

Did you change the turn start speed? Otherwise it is turning when it reaches the set speed and that is well before your set altitude. It turns on the first condition it reaches, altitude or speed. 8km is pretty late to be turning in the new aero and it's probably set that way for auto only as a safeguard since you should hit a proper turn speed well before that. The proper time to start a turn is dependent mostly on TWR so speed is a better generic indicator of when to turn than altitude for unknown craft.

Ahhh. Ok. Then it makes sense that it worked after I had done a complete launch.

No, i didn't change the turn start speed. I guess this is something new, and that is probably my problem.

Thank you for responding so quickly.

LGG

[slaintemaith]

20 post in 6 hours by 4 posters ? You re kidding ?!?

slaintemaith : distant node (month) seems to be unstable in 1.0. Use the warp assistant to move closer to the node and the burn will go fine. I ll see what I can do (most likely warp without aligning if the node is in a long time)

Sorry. I tend to wax loquacious.

I'll try the warp assistant. Never bothered with it before. At least it sounds like a functional workaround for the issues I'm having.

I'll also make sure to post screenies with any future bug reports.

Thanks for being so steadfast and resolute in an atmosphere of extreme pessimism.

[barfing_skull]

Apologies if these have been noted further back than 3-4 pages. Two problems I've seen:

Linux, 64-bit.

1) Occasionally, possibly when the game crashes, some features that were already enabled such as Delta-V, Surface Info, and Orbit info become unavailable upon restart. I'm guessing it's saving something mid-game that gets cleaned up on a normal exit? Completely removing the MechJeb2 directory and re-installing seems to work, but I have to set up my windows over again.

2) In 2.5.1, and build 462, when I use Maneuver Planner, it creates the node, then won't do anything about it other than orient the ship. Auto-warp doesn't work, and it won't execute the maneuver at the specified time. When I go back to build 460, I don't have this problem. I guess to recreate, take a vessel already in orbit, create a node and try and execute it?

Cheers,

-BS

[Jim DiGriz]

Second problem (this may not be a problem): I set everything up and hit "Engage autopilot", and the engines immediately turn on and it takes off. I'm used to it waiting until I hit the space bar to initiate the launch.

I've seen this, but infrequently (and actually not recently, so it might be fixed in dev builds).

Third problem, and this is a doozy: It seems to totally ignore the ascent profile, and starts leaning over starting around 500 meters or so. I tried editing the ascent path, no change. In Automatic Altitude turn, the Turn start altitude is 8.4 km. In manual mode, I set the turn start altitude to 8 km. Didn't seem to make any difference

There's some wobble at liftoff with large rockets that seems to be in the base game. If it wobbles enough and you have AoA limiting turned on then you can get stuck pointing way east and the AoA limit means MJ won't correct it enough. Then with the rocket pointing east once you've built up enough speed and if your rocket has fins and is aerodynamic enough then you're stuck in a really aggressive gravity turn.

Solution seems to be to fly without the AoA limit checked and just make sure your curve doesn't cause AoA to get exaggerated too much and put fins on your rockets so that they don't flip when your gravity turn starts in the lower atmosphere. I use the delta wings as fins on the center stage of big rockets and sometimes one fin on the outside of the boosters. Of course with that much stability then if you do wind up in one of those shallow turns inside the atmosphere then there's really no way out of it.

At least that's what I've been telling myself and the process I've followed seems to avoid the problem.

Edited May 19, 2015 by Jim DiGriz

[GERULA]

We who are about to fly, salute you!

MechJeb, MechJeb, MechJeb..

[barfing_skull]

I've seen this, but infrequently (and actually not recently, so it might be fixed in dev builds).

There's some wobble at liftoff with large rockets that seems to be in the base game. If it wobbles enough and you have AoA limiting turned on then you can get stuck pointing way east and the AoA limit means MJ won't correct it enough. Then with the rocket pointing east once you've built up enough speed and if your rocket has fins and is aerodynamic enough then you're stuck in a really aggressive gravity turn.

Solution seems to be to fly without the AoA limit checked and just make sure your curve doesn't cause AoA to get exaggerated too much and put fins on your rockets so that they don't flip when your gravity turn starts in the lower atmosphere. I use the delta wings as fins on the center stage of big rockets and sometimes one fin on the outside of the boosters. Of course with that much stability then if you do wind up in one of those shallow turns inside the atmosphere then there's really no way out of it.

At least that's what I've been telling myself and the process I've followed seems to avoid the problem.

I'm having the same "MechJeb won't honour my ascent profile" problem. I'm not sure it's related to AoA. I've seen it with every vessel. Plus, the navball guidance changes that early in flight (500-3000m, even when I have it set to 8-10km), so the rocket is just following what MJ is telling it to do via the guidance, without external forces at work. I would think that early deviations from a perfectly vertical flight wouldn't be so much on AoA, but on control itself.

[Jim DiGriz]

I'm having the same "MechJeb won't honour my ascent profile" problem. I'm not sure it's related to AoA. I've seen it with every vessel. Plus, the navball guidance changes that early in flight (500-3000m, even when I have it set to 8-10km), so the rocket is just following what MJ is telling it to do via the guidance, without external forces at work. I would think that early deviations from a perfectly vertical flight wouldn't be so much on AoA, but on control itself.

The initial wiggle seems to have nothing to do with MJ. Try uninstalling it and seeing if you get the wiggle blasting off straight with just SAS turned on or even without SAS turned on. I haven't tried it without the dll loaded, but other than that it seems to be there no matter what I do. Once you've significantly deviated then AoA limiting will cause the initial angle on liftoff to remain fairly large and then once speed builds up you're sunk. Fins on your rocket are also AoA limiters which is what causes the problem once you get moving fast -- you want that in order to avoid flipping, but as a result you have to watch how aggressive your gravity turns start. Also if your initial turn starts at zero and is simply too sharp then the same problem occurs. You want to start the gravity turn soon, but not too aggressively. I've been using a start altitude of 0, end altitude of 55, final angle of 10 and turn shape of 66 fairly successfully (although I just about flipped on my last launch a few minutes ago when I forgot fins, so that profile can probably be better and then likely fins can be smaller).

Anyway, TL;DR i think there's a bug in the unmodded game causing the initial wiggle, and the rest of the issues are all due to the new aero model not MJ.

[cicatrix]

+1 to the Ascent guidance problem

Installed v. 2.5.1.

Had the following parameters:

Turn start alt: 8 km

Turn shape 50%

Limit AoA 7%

MJ started the turn as soon as it get off the launch pad and flipped over the rocket.

Tried several times both with 'Limit AoA' option and without it. Ascent path is ignored as it appears. No, the turn shape parameter is respected since MJ changes the pitch as I slide it left and right, but the Turn Start parameter is completely ignored.

[Carsaxy]

Hi! I'm having some problems with the Mechjeb part.

I found that it often explodes on the launchpad when:

- it is inside fairings (of the mod Procedural Fairings);

- it is touching a part which is not it's parent part.

If it is important: I have Deadly Reentry installed.

I wote it also in the thread of the other mod because I don't know which mod causes the problem.

Thanks

[acc]

I get strange dV/TWR calculations. both is much higher in the VAB than in flight. The only thing I installed/updated was the latest deadly reentry.

data:

- KSP 1.0.2 x86

- MJ dev #459

- DR 7.0.3

- log

- installed mods

also reportet at DR thread

Edited May 19, 2015 by acc

[Galane]

Hi! I'm having some problems with the Mechjeb part.

I found that it often explodes on the launchpad when:

- it is inside fairings (of the mod Procedural Fairings);

- it is touching a part which is not it's parent part.

If it is important: I have Deadly Reentry installed.

I wote it also in the thread of the other mod because I don't know which mod causes the problem.

Thanks

Get rid of deadly reentry. Parts overheating and exploding is now in the stock game and it can be adjusted or turned off when you start a new game. (Or if you want to dig into the save and persistent files you can alter the settings there.)

If you're using 1.0 or 1.0.1, update to 1.0.2.

[sarbian]

acc : #459 had a wrong pull of the dv code. Current dev should work.

Carsaxy : most likely a heating problem with the part. Strange that it occurs inside fairings. I'll test with DRE but not today.

barfing_skull / Jim DiGriz : the problem is that the vessel move a bit sideway on launch and MJ correct a bit too much. I may have to add a 1/2 second where the AP lock control before starting to pilot.

And if I get one more question about the damn ascent path I won't post for a week. I know the thread is moving fast but read it few page at least.

[acc]

acc : #459 had a wrong pull of the dv code. Current dev should work.

alright, thank you. but I tracked it down to procedural heatshields. it caused the huge difference in numbers. maybe the dev version of MJ was calculating wrong at all, but the p. heatshield messed it up. I tried to use them (stupid me) as structural landing feets.

Javascript is disabled. View full album

(first two images with DR .65m shields, another two with .65m p. shields)

Edited May 19, 2015 by acc

[Azazael]

I am running KSP 1.02 and MJ 2.51.

Assent autopilot works fine now with no more toppling over with large rockets,

Rendezvous works well too but I cant get the docking autopilot to work.

What happens is that it aligns perfectly with the docking axis but its a whole bounding box out (in my case 87m) the other side of the docking port. Is this a known issue ?

[RuBisCO]

I'm having a problem with 2.5.1: sometimes it refuses to do burns, it just sits there and does not auto-warp or burn, it does aim at the maneuver node.

[cicatrix]

I am running KSP 1.02 and MJ 2.51.

Assent autopilot works fine now with no more toppling over with large rockets,

Rendezvous works well too but I cant get the docking autopilot to work.

What happens is that it aligns perfectly with the docking axis but its a whole bounding box out (in my case 87m) the other side of the docking port. Is this a known issue ?

Did you select your docking port and choose 'Control from here'?

[sarbian]

I can't fix that node problem until someone give me steps to duplicate it. Yesterday I tested it with a kerbal X and the launch went fine, the circularization went fine and the burn to duna went fine too.

[linuxgurugamer]

I'm having the same "MechJeb won't honour my ascent profile" problem. I'm not sure it's related to AoA. I've seen it with every vessel. Plus, the navball guidance changes that early in flight (500-3000m, even when I have it set to 8-10km), so the rocket is just following what MJ is telling it to do via the guidance, without external forces at work. I would think that early deviations from a perfectly vertical flight wouldn't be so much on AoA, but on control itself.

For me, the problem was that the ascent profile started when the speed was hit, which was pretty fast for the rocket I was using.

My solution was twofold:

1. Increase the speed in the Ascent Profile where it would start to lean

2. Reduce the acceleration of the rocket (it was pretty overpowered)

LGG

@sarbian, ignore any messages about the ascent profile for now. I'll see if I can keep my eye on this thread and respond as necessary to those, since it is actually working properly, people just don't understand the new stuff. and thanks for all the work you do on this and other mods

[Ninjahgoblin]

I made an account to post feedback. 1st- Thank you for this outstanding mod. Has helped me enjoy the game worlds over.

ISSUE:

-KSP 1.0.2

-Mechjeb 2.5.1

Every single time I use ascent guidance, no matter how I fidget with the turn start or shape. It starts the Gturn at around 1000, and proceeds to start doing flips after10000ish. Never getting an Apoapsis above 20k.

I am not sure what to do, I have tried the "more fin method" I have tested using designs I know fly into orbit. Still no ascent.

Do you need any more details? Rig info etc.?

[Ninjahgoblin]

For me, the problem was that the ascent profile started when the speed was hit, which was pretty fast for the rocket I was using.

My solution was twofold:

1. Increase the speed in the Ascent Profile where it would start to lean

2. Reduce the acceleration of the rocket (it was pretty overpowered)

LGG

@sarbian, ignore any messages about the ascent profile for now. I'll see if I can keep my eye on this thread and respond as necessary to those, since it is actually working properly, people just don't understand the new stuff. and thanks for all the work you do on this and other mods

Thank you for this. I read a small passerby snip of someone else stating the turn is starting early because it hits the Speed variable 1st before the height variable.

Will test this when home and post back. But I believe this is the fix.

Sorry for posting the same issue as many others!

[tru7h]

Hi Sarbian,

I've read the past dozen or so pages and I might have a bit of information regarding the maneuver node bug. I was using DEV build 460 of the mod yesterday and was having an issue with the maneuver node not being triggered. It appeared that having two mechjeb units on my rocket (one on my probe and one on a recoverable booster stage) was causing issues - any node created when they were still coupled together would fail to execute but once the stages were decoupled, all maneuver nodes would execute flawlessly.

I am currently at work and unable to provide any logs but I tested this repeatedly and only noticed the issue when there were more than one MechJeb units present on my rocket. I never had this issue before starting to use the DEV builds this past weekend.

I hope it helps and I'd like to echo the praise for you and your mod - KSP just isn't the same without it! Thank you ad infinitum

[sarbian]

Before I let linuxgurugamer take over the replies for the ascent AP let me explain : this change was made so the default profile works better with the 1.0 changes. The old method of going straight up for a few KM and then turn hard is not optimal anymore and will result in flip most of the time. With the early turn you don' t get a sudden change of the AoA and your vessel should not flip.

This and the limit AoA default (and you may not have those default if you have old MJ configs files) should worked fine with all the vessel I tried (which is of course a small subset of what users will use).

Most of the old KSP tutorial about ascent are not relevant since 1.0 and if you follow those you ll get problems. Or use tutorials intended for FAR, the new system is not that different for rockets.

tru7h : I missed your post. Nice, I hope it is just as stupid as that (even if I have no idea how 2 MJ unit could be active on the same ship)

Edited May 19, 2015 by sarbian

[tru7h]

tru7h : I missed your post. Nice, I hope it is just as stupid as that (even if I have no idea how 2 MJ unit could be active on the same ship)

Me neither. As I said, I had never noticed this issue before this weekend either. If it helps, I'm using Linux x64 and MechJeb/Engineer for All (although I did try with and without the actual parts - same deal).

Thanks again for all your work

[Ninjahgoblin]

@Sarbian

My issue with the below in bold. Is that yes, the turn starts early and I don't mind. But my apoapsis never gets above 20km. Ever. The turn starts early, I reach an exceptionally low apoapsis (when successful), I burn up in atmosphere while trying to complete Gturn.

Also with the Limit AOA default setting of 5', it always goes past this and does a hard 90' turn from about 20sec after take off

I have done reinstalls, fresh isntalls, just Mechjeb and KSP, the default settings and playing with the Edit Ascent path variables.

Maybe I am not understanding things completely.

Before I let linuxgurugamer take over the replies for the ascent AP let me explain : this change was made so the default profile works better with the 1.0 changes. The old method of going straight up for a few KM and then turn hard is not optimal anymore and will result in flip most of the time. With the early turn you don' t get a sudden change of the AoA and your vessel should not flip.

This and the limit AoA default (and you may not have those default if you have old MJ configs files) should worked fine with all the vessel I tried (which is of course a small subset of what users will use).

Most of the old KSP tutorial about ascent are not relevant since 1.0 and if you follow those you ll get problems. Or use tutorials intended for FAR, the new system is not that different for rockets.

tru7h : I missed your post. Nice, I hope it is just as stupid as that (even if I have no idea how 2 MJ unit could be active on the same ship)