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

[RyanK]

Arghh. Frustrated. Using the 466 build. Still no luck getting MJ to activate or control the throttle at node execution. New windows 7 install with new KSP install and all the latest versions of the mods. Any ideas?

Log

Mods

Deleted all other mods. Still not working properly. Attached latest log file. Totally stumped.

Log

[RyanK]

Deleted all other mods. Still not working properly. Attached latest log file. Totally stumped.

Log

Well, discovered that there is no problem when playing in sandbox mode with a stock kerbal ship. In my Science campaign game the engines will not fire at the nodes. Any ideas? Something wrong in the science tree?

[Two]

So I gave this mod a try today, and if I didn't knew it any better (from all the high praises around here), I would say this is a crap mod. I am not sure if the current version is just buggy or if I do something wrong big time, but half of the features failed to execute either completely or properly.

I tried the ascension assist with 5 different rockets, and either MJ just flipped the rocket around then smashed it into the ground at full acceleration (despite all edits I made in that module), or it got into space, missed the node to orbit and waited patiently till the rocked smashed into the ground again. Even after I steered the rocket into orbit manually (simply by auto-hold prograde, then wait, it's not a difficult rocket to fly at all!), MJ refused to adjust orbit and was stuck on "Retracting solar panel", which are not retractable. Also auto-staging just blew through all the stages for no particular reason, shredding away the protective air-shell on top at 2km height.

Once I steered the rocket into 80km orbit manually I tried to fly to Mun, using the advanced transfer. Which simply failed to do anything. Something flickered at the bottom of the window, but no node was added or anything else. So I tried the Hohmann-Transfer... which failed as well without any error message. So I used the standard transfer, which actually plotted a suicide curse to smash into the moon. I couldn't find any module to correct that, so I once again manually adjusted to get a 20km periapsis.

Circularizing then landing however did work properly, so at least something.

[smjjames]

You should at least post an output log or two (or three or four) if you really want Sarbian to help you.

On the AP getting stuck on retracting panels, I reported that one earlier, and you could have just used maneuver planner to circularize at apoapsis anyway.

Some screenshots of however the heck you're setting the ascent AP would be good.

Edited May 31, 2015 by smjjames

[Two]

Where would I find that output log? The MechJeb2.log is basically empty.

I tried many different ways on ascent AP for over an hour with various rockets from small to big, including all options on and off and even an adjusted ascent curve. MJ couldn't get a single rocket into orbit properly. I was able to fix the over-steering with the AoA setting combined with 'limit to terminal velocity', however in that case MJ always began turning the rocket for AP as low as 40km, then missed the maneuver every single time and crashed into the ground. My guess is that the drag caused by the early turn - which makes you loose >1km AP - caused the ascent path to miss the node, and MJ isn't adjusting to that properly.

[BudgetHedgehog]

Where would I find that output log? The MechJeb2.log is basically empty.

It's a .txt file in the KSP_Data folder.

[Two]

I found several MechJeb related things in that file:

http://pastebin.com/G1h7wyDE

Edited May 31, 2015 by Two

[smjjames]

Where would I find that output log? The MechJeb2.log is basically empty.

I tried many different ways on ascent AP for over an hour with various rockets from small to big, including all options on and off and even an adjusted ascent curve. MJ couldn't get a single rocket into orbit properly. I was able to fix the over-steering with the AoA setting combined with 'limit to terminal velocity', however in that case MJ always began turning the rocket for AP as low as 40km, then missed the maneuver every single time and crashed into the ground. My guess is that the drag caused by the early turn - which makes you loose >1km AP - caused the ascent path to miss the node, and MJ isn't adjusting to that properly.

Hm, are you using FAR? Because MJ needs a dll extension to help it understand how FAR works and that isn't updated atm. Sarbian did say he would look into it. If you're using FAR, that could explain why it's failing with the ascent.

Edit: You're supposed to post the whole thing.

[Two]

Hm, are you using FAR?

Currently using KIS, KER, Simple Science Fix and Stage Recovery.

[futrtrubl]

I'm having some trouble with mechjeb not adding in the angle to launch to rendezvous box. Normally it adds in the angle, but out of nowhere it refuses to work. I can add in my own angle, but without being too accurate, it's worthless, and I made my rockets to barely scrape by with the necessary Dv to do their jobs, because I feel that extra fuel is a waste of money. Any help would be great, cuz for all I know mechjeb has become sentient and has begun phase one of the invasion of kerbin!

Have you already made a normal ascent to the rendezvous altitude? I mean completely including circularization burn? If not then mechjeb has no way to know what phase angle it should launch at.

- - - Updated - - -

might have been answered already and i havent seen it..but i DLed the newest version of MJ and my ascent guidance window is gone. no options for it no nothing. is it farther along on the tech tree now? im not seeing it listed if it is. anyone have any ideas how to remedy this?

Have you upgraded your tracking center sufficiently? It can't give you the ascent guidance, or most things, if you can't make maneuver nodes etc.

[vardicd]

Don't know if its ever been suggested before, or even if its possible, but here's a suggestion I'd like to have for MJ. :

[gilflo]

I think there is definitely something wrong with the indicated weight of a lot of stuff. But I dont know if it is KER or KSP parts or Mechjeb

Here is a Mk2 cockpit (2,06T) , then i add a MK2 rocket fuel fuselage (3,57T) and a Rapier Engine (2T)

Total weight should be 7,63T

In fact it is 7,435T, because if you look the weight of the fuselage (right click) wet weight given is 3,375T and not 3,57T given in SPH Fuel tank part.

Then I add 2 delta wings. Given weight of each is 0,2T as indicated in aerodynamic parts. Total weight should be 7,435+0,4 = 7,835T

In fact it is 8,279 . So each delta wing weight is in fact 0,422 and not 0,2T

If you add mechjeb on the aircraft, it will find the same total weight as KER. They agree.

And you can try a lot of wings, each time the real calculated weight is not the indicated weight.

So there is something wrong somewhere with the weight and in final, my SSTO is given for 67T by KER and 57T by Mechjeb.

Here Mechjeb is right and disagree with KER

But when i launch my SSTO, Mechjeb take the wrong KER weight in account, 10T more! Weight is now 67T

In other way, I can see that mechjeb don't take in account weight of some small parts, when I add them, like vernor engines or adjustable landing gear......

[smjjames]

Does that discrepancy still show with the latest dev build?

edit: I notice that you said in the KER thread that you have FAR, so uh, no idea since I don't use the new version of FAR.

Edited June 1, 2015 by smjjames

[gilflo]

I think i have the latest one. The difference between Mechjeb and KER should be around 3T, according to the wings weight which is different when using FAR.

I built the same SSTO again and know final weight is 61T which is correct. I don't know why the previous one was 67.

[sarbian]

Your screenshots are not from a dev build (or I broke the version display again but I doubt it)

[The Lone Wolfling]

The porkchop plot doesn't seem to be working (dev build 2.5.1.0-466):

[EXC 12:52:32.407] NullReferenceException: Object reference not set to an instance of an object
	MuMech.OrbitExtensions.SwappedOrbitNormal (.Orbit o)
	MuMech.OrbitExtensions.RelativeInclination (.Orbit a, .Orbit 
	MuMech.OperationInterplanetaryTransfer.MakeNodeImpl (.Orbit o, Double UT, MuMech.MechJebModuleTargetController target)
	MuMech.Operation.MakeNode (.Orbit o, Double universalTime, MuMech.MechJebModuleTargetController target)
	UnityEngine.Debug:LogException(Exception)
	MuMech.Operation:MakeNode(Orbit, Double, MechJebModuleTargetController)
	MuMech.MechJebModuleManeuverPlanner:WindowGUI(Int32)
	UnityEngine.GUI:CallWindowDelegate(WindowFunction, Int32, GUISkin, Int32, Single, Single, GUIStyle)

This is with an asteroid tug that I keep having problems with (of the random rotation / acceleration leading to explosions variety) - perhaps that has something to do with it?

[Val]

...

Then I add 2 delta wings. Given weight of each is 0,2T as indicated in aerodynamic parts. Total weight should be 7,435+0,4 = 7,835T

In fact it is 8,279 . So each delta wing weight is in fact 0,422 and not 0,2T

...

FAR changes the weight of wings depending on the Strength slider. If you didn't change the strength, then it seems the default setting is a lot heavier than Stock weights.

[mikegarrison]

Arghh. Frustrated. Using the 466 build. Still no luck getting MJ to activate or control the throttle at node execution. New windows 7 install with new KSP install and all the latest versions of the mods. Any ideas?

I have occasionally had a problem where MechJeb will not autowarp to a node nor burn the node. In fact, I can't even make the burn manually unless I delete the Mechjeb node first. However, I think it may have been caused by some funkiness due to docking/undocking. My current suspicion is that after undocking, for some reason the engines are no longer staged (even though they are activated), and this seems to be confusing Mechjeb.

[sober667]

Ummm i have some problems i dont now why i bet that is mechjeb but not shure... smart A.S.S. not working so good if i say to keep any direction its mess things up(spinn out of control or hold prograde veven if its target mode )

i have remotech and FAR and litlle bit of other mods

[Capt. Hunt]

Ok, I'm still having issues with ascent guidance. It can't be my rocket designs, because I'm giving them plenty of extra fins and they launch just fine if I use version 2.5.0 mechjeb (the last pre 1.0 build), but they're all over the sky if I use the current build of mechjeb. Before you ask, this is all with 1.0.2 KSP.

With current MJ it's almost like the guidance can't tell what the rocket's actually doing and is just sending the controller inputs on a timer. On some of my launches, the rocket veers off course and instead of correcting it to 090 MJ just continues on the same trajectory as if nothing happened. On other launches I've had it make orbit way off altitude and then the circulization burn ends up offset by whatever the the altitude error is. I'll set it to 75 km before launch, the rocket goes to 100 km and then circularizes to an orbit with an apokerb of 100 and a perikerb of 50.

This is all very frustrating, isn't it reasonable to expect that if a rocket can launch with a version that is not optimized for the new aero it should work with a version of MJ that is?

The only reason I made the upgrade at all was I noticed that landing guidance was a bit wonky in the older version (which I assumed was due to the new aero. Unfortunately as I can barely launch with the new MJ I have yet to be able to if LG works any better.)

Edited June 1, 2015 by Capt. Hunt

[skips]

.... I'll set it to 75 km before launch, the rocket goes to 100 km and then circularizes to an orbit with an apokerb of 100 and a perikerb of 50. ...

This result should be expected. You asked for a major axis of your orbit to be kerban's radius + 75 km. MechJeb provided that semi-major axis. The issue is why your rocket failed to stop accelerating when the apoapsis reached 75 km. It may have to do with the other mods that you have installed, because my rockets generally hit the desired apoapsis within 1-5 km for requested altitudes between 75 and 300 km.

skips

[mikegarrison]

This result should be expected. You asked for a major axis of your orbit to be kerban's radius + 75 km. MechJeb provided that semi-major axis. The issue is why your rocket failed to stop accelerating when the apoapsis reached 75 km. It may have to do with the other mods that you have installed, because my rockets generally hit the desired apoapsis within 1-5 km for requested altitudes between 75 and 300 km.

skips

No, what happens is not that the rocket keeps accelerating after reaching an ap of 75km. What happens is that if you have low thrust/weight and you try to circularize at ap, you end up with a pretty elliptical orbit because half your burn happens when you are dropping back down again. If you drop all the way back to 50 km before you finish the burn, you are obviously going to end up with a peri of 50km.

[Capt. Hunt]

ok, maybe that one was a fluke, it also tends to finish 5-10 km low of my target orbit too, before circularizing. At this point less then one in ten of my launches even gets that far. I'm more concerned about the behavior right off the pad. Why's a rocket that works fine with the old mechjeb (and 1.0.x aero) suddenly not working when I upgrade mechjeb? If anything I'd expect the opposite to be true; with the new aero, rockets should be performing better under control of the new version of mechjeb.

Edited June 1, 2015 by Capt. Hunt

[mikegarrison]

ok, maybe that one was a fluke, it also tends to finish 5-10 km low of my target orbit too, before circularizing. At this point less then one in ten of my launches even gets that far. I'm more concerned about the behavior right off the pad. Why's a rocket that works fine with the old mechjeb (and 1.0.x aero) suddenly not working when I upgrade mechjeb? If anything I'd expect the opposite to be true; with the new aero, rockets should be performing better under control of the new version of mechjeb.

I too have had problems right off the launch pad. Mechjeb ascent control seems to not control the rocket's direction at all until it tries to start to turn, and by then the rocket may be pointed off in some crazy direction. With the same rocket under SASS control (just a simple "kill rot surface" mode), it goes up perfectly straight until I tell it to start rotating.

(I should add I am not using a development build of Mechjeb, I am using the 2.5.1 released version.)

[skips]

ok, maybe that one was a fluke, it also tends to finish 5-10 km low of my target orbit too, before circularizing. At this point less then one in ten of my launches even gets that far. I'm more concerned about the behavior right off the pad. Why's a rocket that works fine with the old mechjeb (and 1.0.x aero) suddenly not working when I upgrade mechjeb? If anything I'd expect the opposite to be true; with the new aero, rockets should be performing better under control of the new version of mechjeb.

My best guess is that you have AOA enabled and when it cuts off your thrust because the aerodynamics are overpowering your attitude control, the center of lift changes significantly and your attitude control cannot cope. Try turning off the AOA limiter and adjusting the shape of your ascent profile to better match the changing direction of your velocity. Ideally you should be able to adjust your ascent profile to match the natural gravity turn of your craft and never trip the AOA limit, but that takes a significant amount of fiddling with the profile and your thrust limiter. BTW: none of these issues are directly MechJeb problems.

skips