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

[Galane]

MechJeb works OK with Stock Drag Fix, which mainly eliminates the mass of fuel tank contents being added into drag calculations. (Other features TBD.)

No problems on ascent though rockets that would hit terminal velocity don't with SDF yet accelerate faster and reach orbit with more fuel left.

On landing (tested on Kerbin and Eve) on bodies with an atmosphere, MechJeb doesn't make a long enough high deorbit burn. It goes long but then aimes the nose retrograde and down to do a long correction burn. From there it handles the rest of the descent normally - though you get less deceleration from drag. May have to revise parachute arrangements (especially with deadly reentry) and find somewhere to put more fuel if your lander is built to hit the dirt on fumes with stock aerodynamics.

If you are staging things off the top end of your rockets you will absolutely need moar boosters to kick them up and away due to the higher speeds the rockets will hit by the time the empty tanks have to drop. I resorted to surface attaching a pair of slimmed down stretchy SRBs to each largest Kerbodyne tank, with the thrust kicked up to around 1700 with a 4 second burn time. Just a *bit* overkill because they'll shoot ahead well over 1,000 meters.

If you're tired of the "soup" but don't want to go to FAR or even NEAR, try SDF http://forum.kerbalspaceprogram.com/threads/84978-0-24-x-Stock-Drag-Fix It won't break things.

[Wanderfound]

Incidentally, "force roll" seems to think that 0° is inverted while suborbital in vacuum, but normal while in atmosphere or orbit. Is this intended? Not really a problem, but it does waste a bit of RCS.

[sarbian]

wobble fixed in #308. I was adding the wrong value and we ended up with twice the engine torque.

As for the force roll I did not write the patch but it looks like it's intended

[jebandhisfriends]

hey guys , my mechjeb doesnt work at all i installed the latest version in 23.5 cause the previous one stopt working without a reason. but now my ksp is telling me that its incorrect path and if i changes that it still sent that message so what is the problem, can you help a little maybe?

[FlexGunship]

wobble fixed in #308. I was adding the wrong value and we ended up with twice the engine torque.

As for the force roll I did not write the patch but it looks like it's intended

#308 either has a new feature I'm unaware of, or has an additional bug. Since changing to #308, I am no longer able to see more than on conic patch. That is to say, if I have an encounter with the Mun, I get my primary (blue line) and nothing else. There is no secondary or tertiary conic showing the result of an SOI change.

I went back to #297 (my go-to version) and my conic patches were back to normal. I added two "part only" mods that I did not attempt to uninstall for this test, so I apologize if this is unrelated to your update in #308.

EDIT: To be clear... I did NOT test #305 for this. So I don't know how long it has been present.

DOUBLE EDIT: Checked the KSP setting file and confirmed that the CONIC_PATCH_LIMIT is still set to 3. I uninstalled the two mods I had added. This problem does not exist in #297, but shows up when I use #308. So, I believe this bug was introduced between those two revisions.

hey guys , my mechjeb doesnt work at all i installed the latest version in 23.5 cause the previous one stopt working without a reason. but now my ksp is telling me that its incorrect path and if i changes that it still sent that message so what is the problem, can you help a little maybe?

Current version of KSP is 0.24.2. The current version of Mechjeb is compatible with that. Either download an old version of MechJeb or get an updated version of KSP.

Edited August 31, 2014 by FlexGunship

[cocowawa]

Hello guys, i have a problem when trying to make Mechjeb's interfaces to work on version 0.24.2.

Maybe the problem its that im doing something wrong, but i copied the uncompressed folder, downloaded from this thread, of MechJeb2, inside the KSP/GameData folder. But nothing in the game changes except that i find a device on the "control" tab of the Vehicle assembly. But i can't install it on any vessel, there is a message "Requires an entry purchase in r&d."

Well.. i have used the search tool of this forum and it took me to a thread, with this solution :

To All, who don't know how to Purchase the Mobile Processing Lab, i know how to get it:

- Open the Research Center

- Search for the Category Space Exploration

(It is in the middle of the Tree, from left to right: Start -> Basic Rocketry -> Survivability -> Science Tech -> Space Exploration)

- Click on it

- You should see at "Availabe Parts" right side on Screen, from left to right: Pegasus I Mobility Enhancer -> 2HOT Thermometer -> Mobile Processing Lab MPL-LG-2

- Left-Click on the new MobileProcessingLab-Picture and it opens a Context-Menue

- There Click Research

I have installed the Zip'ped Version of the Game and i have it

I hope i could Help

When i try to follow the steps, i found out i already own such parts, so i can't right click'em.

So i also try to right click to open the more info menu on the item, in the vehicle assembly menu, but it hasnt context menu.. i cant find any "Research" button or link to click.

What i want about mechjeb its that interface to know the degrees between kerbin and other planets, that interface would help me a lots...

Can someone help me?

Thanks!!

[chucksterock]

Mechjeb noob here. Is there a walk thru or tutorial or something available for the new version? Some steps outlining how to program a launch and landing would be nice. TIA

[drtedastro]

Hello to all. While i am certainly loving all of the great new layouts / additions to MJ, i am now unable to get the docking to work in about 90% of attempts. There 'seems' to be something regarding where the 'docking' vessel is in relation to the object to be docked to. I have read some other posts that seem to be having same issue(s). I was trying #308, but am now dropping back to find 'stable' docking version. Any ideas, comments, etc would be helpful.

Best to all and thanks.

[Guest]

Not a MechJeb newb here. Running #308 and am quite happy. Ascent, docking, landing, all that seems to be working as expected. However, on my heavy lift vehicles (think Saturn V), the yaw seems a bit overactive. As in the rocket fishtails after the gravity turn and atmospheric pressure becomes negligible. Let me know what information you require and how to gather it for you, and I'll be happy to comply.

[Ippo]

Hey guys, cross-posting fom the RO thread: is there any known issue with mechjeb in Real Solar System + Realism Overhaul?

It seems to be acting up in my install: it ignores Smart ASS commands, it doesn't execute the gravity turn at all, when using the docking autopilot it keeps getting *away* from the target until I disable it.

Does it ring any bell, by chance?

[daawgees]

Not a MechJeb newb here. Running #308 and am quite happy. Ascent, docking, landing, all that seems to be working as expected. However, on my heavy lift vehicles (think Saturn V), the yaw seems a bit overactive. As in the rocket fishtails after the gravity turn and atmospheric pressure becomes negligible. Let me know what information you require and how to gather it for you, and I'll be happy to comply.

If I'm understanding your issue correctly I was having the same problem as well. On the 'Ascent' window disable 'corrective steering.' My wording may be a tad off as I'm not running KSP right this second, but turning that setting off during ascent helped me.

[BARCLONE]

Not a MechJeb newb here. Running #308 and am quite happy. Ascent, docking, landing, all that seems to be working as expected. However, on my heavy lift vehicles (think Saturn V), the yaw seems a bit overactive. As in the rocket fishtails after the gravity turn and atmospheric pressure becomes negligible. Let me know what information you require and how to gather it for you, and I'll be happy to comply.

I'll toss this in from my observations...

On the AGAP, there is a checkbox for "Limit acceleration to", and the default value is 40m/s/s. Try checking this box, but lower the value to something around 15 m/s/s. Basically, just enough to get the rocket moving off the pad, not jumping off. If the rocket's acceleration is too high, MJ has to fight harder to keep the rocket straight. Also, start the turn as close to the ground as possible. This will depend on the rocket, of course, but I'm having reasonable success with the turn starting at 1 Km. For getting off Kerbin, set the curve number to around 55%.

You want the level-out altitude to be just as it reaches the edge of the atmosphere, so a value of 68-75 Km is fine. Trying to set the orbit too low can also cause problems. Even if your target altitude is in the 75-80 Km range, like for some of the contracts, try setting a higher altitude initially then adjusting back down with Maneuver Planner.

[daawgees]

I'm having a problem of my own now. Info first: KSP .242 32-bit, Windows 8.1 x64; using MechJeb 308. The issue I'm experiencing involves Landing AP; I've noticed that during descent to any surface the AP fires the engines in spurts rather a nice steady slow burn to the ground. This results in my craft coming to a hover several hundred meters above the landing point and remaining there while the ship turns about firing the engines in bursts. In order to rectify this I did try to turn on the 'smooth throttle' option under utilities but this has not worked. I'm sure there is further information that I need to provide, anyone have any idea about how to resolve this?

[Bothersome]

Hey guys, cross-posting fom the RO thread: is there any known issue with mechjeb in Real Solar System + Realism Overhaul?

It seems to be acting up in my install: it ignores Smart ASS commands, it doesn't execute the gravity turn at all, when using the docking autopilot it keeps getting *away* from the target until I disable it.

Does it ring any bell, by chance?

Yea, I'm using build 308 also and notice it's not as good as the release version, especially for bigger crafts. In RO, it is critical to stay within the gravity turn profile and that profile needs be close to where the nose is pointed into the wind. Anything more than 5 degrees off below 25km is asking for trouble.

Some of the figures I've been toying with are in the attitude settings, turn off auto-adjust and set the parameter at the top to 0.35 to 0.5. The lower the number the tighter MJ tries to keep it. However in 308 there seems to be some kind of delay or chasing the path action going on. I know the rocket is built well because SAS can hold the angle very well without shaking or chasing. Yet when MJ 308 is running on "use stock SAS" it's some kind of terrible.

Also in RO, I found using a 1k start gravity turn and 60% turn profile with the end of the turn at 240km. Rockets should have a start TWR of 1.30 and have a TWR of about 2.0 by the time it gets to 20km. The TWR should not exceed 7.5 at any time lest you risk breaking something. So you need to set your thrusts to max TWR (7.5 at empty tank weight) and then load up as much fuel as you can until TWR is at 1.3 for lift-off. That the first stage, other stages you can get by on less starting TWR but you do need to get moving once you get into space, I personally like TWR > 1.0 climbing to 7.0 for upper stages.

[sarbian]

Yet when MJ 308 is running on "use stock SAS" it's some kind of terrible.

When "use stock SAS" MJ does not do anything about attitude. It tells the stock SAS to "go that way" and nothing more. The stock SAS is a bit better inside the atmo but you can tune MJ to do as well.

The problem with MJ in atmo is that Tf gets too low and the controls become over reactive. I need a function that increase the lower limit of Tf when dynamic pressure increase. Not exactly hard to code but hard to tune.

[Haiyooo]

So is the no joy in getting mechjeb to work on 24.2 ?

Downloaded the latest verion of jeb saying it worked on 24.1 but all i get is the modules and no overlay even when attached etc.

[Neutrinovore]

When "use stock SAS" MJ does not do anything about attitude. It tells the stock SAS to "go that way" and nothing more. The stock SAS is a bit better inside the atmo but you can tune MJ to do as well.

The problem with MJ in atmo is that Tf gets too low and the controls become over reactive. I need a function that increase the lower limit of Tf when dynamic pressure increase. Not exactly hard to code but hard to tune.

Yes, I've discovered this as well. Although for me, it's fairly straightforward: When flying a spaceplane in the atmosphere, I can just select 'Use Stock SAS' and the plane flies beautifully. Then when I get out into the vacuum of space, I unclick that and use either 'Tf auto tunning' (note the bold text, that's a typo that's STILL in MechJeb ever since you introduced the auto-tuning option, btw. Just saying. ), or sometimes I have to fiddle with manually tuning the Tf if my ship is too big or small or whatever MJ doesn't like about it that makes auto-tuning not work very well.

Anyway, just thought I'd add my 2 cents. Later:

[sarbian]

lies, there is no such typo !

[drtedastro]

Damn, you are QUICK........

[Starwaster]

There. Are. Four. Typos!

[dtoxic]

KSP 0.24.2 x32

win7 sp1 x64

Mechjeb 309 dev build

got this error while trying to circularize orbit

ArgumentOutOfRangeException: Argument is out of range.

Parameter name: index

at System.Collections.Generic.List`1[unityEngine.Quaternion].get_Item (Int32 index) [0x00000] in <filename unknown>:0

at MuMech.VesselState+EngineInfo.AddNewEngine (.ModuleEngines e) [0x00000] in <filename unknown>:0

at MuMech.VesselState.AnalyzeParts (.Vessel vessel, MuMech.EngineInfo einfo, MuMech.IntakeInfo iinfo) [0x00000] in <filename unknown>:0

at MuMech.VesselState.Update (.Vessel vessel) [0x00000] in <filename unknown>:0

at MuMech.MechJebCore.FixedUpdate () [0x00000] in <filename unknown>:0

(Filename: Line: -1)

https://www.dropbox.com/s/v1nbixxdtjutkkc/MJ-Error.rar?dl=0

[Galane]

So is the no joy in getting mechjeb to work on 24.2 ?

Downloaded the latest verion of jeb saying it worked on 24.1 but all i get is the modules and no overlay even when attached etc.

Get the latest dev build, linked in (too small a font) in the first post.

[Bothersome]

I'd just like to say that MJ 309 is working pretty good for me again. Here is my most complicated rocket so far that MJ can put into orbit for me. I'm using RO (FAR, RSS, DRE, etc). The payload is a 15 ton module from Station Science. The rocket started out very spegetti like, wanting to wobble even sitting on the launch clamps. After securing the rocket more, and adding some fins at the top to help push the nose into the desired location, MJ had no more problems with getting it to orbit. The rocket is a bit under powered below 45km and it starts sliding in the wind a bit. But MJ done a fine job of compensating.

My current mod list for the curios.

Generic Mod Enabler - v2.6.0.157

[D:\Steam\steamapps\common\Kerbal Space Program\MODS]

RealSolarSystem7.2

Advanced Jet Engines

DeadlyReentry

RealChute1.2

RealFuels

TweakScale

ProcFairings

RemoteTech

FinalFrontier

Interstellar

PorcParts

EngineIgnitor

StationScience

MechJeb2

FerramAerospaceResearch

NovaPunch

RealismOverhaul

[Morlock]

Sarbian, if you allow me I can give you an idea of as doing Tf's active autotuning of different form:

1-Supposing that the rocket is a rigid body, and you know the value of the torque and of the moment of inertia, it is possible to estimate (or simulate) the response ( giro speed or position) that should have the rocket in the following delta-time.

2-Then it will be possible compare the response estimated with the obtained one really. Only two motives happen to me for those who are not equal:

a) There is a mistake in the ratio torque / moment of inertia.

The rocket does not behave as a rigid body, does it as a flexible body.

3-If the ratio torque / moment of inertia is correct then the value of Tf must increase (this will reduce the difference between the calculated and real response). If the difference is void then Tf's value will have to diminish looking for the ideal value for Tf.

I admit that to do this work is very complicated and to be possible that it works badly.

Edited September 2, 2014 by Morlock

[sarbian]

We already do that Morlock. Available torque, MoI and a bunch of other stuff is computed every physic frame. And no KSP rocket are not a rigid body, even if each of the part is.

AFAIK the MoI is right. If was consistent with test I made last year when I made my first contribution to MJ code.

Torque is mostly right. The part I am not sure about is the torque from a reaction wheel far from the CoM. Torque from RCS and control surface looks good, but I never did extended test (a bit harder to test when in atmo). Torque from engine was great and should be even better with the last few patch.

The current Attitude controller use all that extensively

But even if you know all the variable you still need a PID controler, and need to tune it. That's where our Tf comes in, as a single variable used to tune the PID controler that works quite well.

As for the current tf autotuning it uses the MoI and the Torque. And I want to add the dynamic pressure somewhere so MJ is less violent with control inside the atmo.

If you have a patch for any of this please submit it.