HOW TO be accurate

[mockingking]

I\'ll be brief.

I\'m using MechJeb to set up sattelite networks, orbital construction sites and ICBM strikes. But my rockets\' accuracy is horrible. Just plain painful to see. Inclination is always off and makes my orbits look like a cat played with them. Altitude can vary from the original design by as much as 20%.

And that\'s without any accidents - the rockets start, stage and fly fine.

So I have no idea what\'s wrong with them. Do I need more winglets (Whole wings?) for the atmospheric phase? Would hooking MechJeb up with some RCS do the trick? Do my rockets need a lot of SAS? Or do they simply need to be as solid as possible; ideally very small, single-stage rockets?

Help, please.

[Endeavour]

Do you use any non-stock parts. I\'ve noticed that mechjeb doesn\'t always work perfectly with mod parts.

[Capt'n Skunky]

Are you making sure to change your ascent path final alt to match your desired alt? You have to set that in both locations or weird things can happen.

Also, make sure you have no SAS or ASAS parts on the craft, they are redundant with MJ and can cause it to get wonky.

There will always be a little variance in your actual flight, but not as great as you seem to be having.

And yes, MJ needs some RCS for control. I usually place 4 at the base of each stage fuel tanks just above the engines. Use the single jet versions(C7 design), not the 4-way thing. I only use the 4 way on the lander/payload.

Cheers!

Capt\'n Skunky

KSP Community Manager

[BlackBicycle]

Can\'t tell for sure without looking at your vehicle, but I have a theory:

Maybe your rocket design is sub-par. Could you post a picture or two of the thing you are trying to put in orbit? Even mechjeb cannot help if the rocket is poorly designed. Lack of atmospheric control surfaces, failure to (provide and) switch on the RCS when in upper atmosphere and a few other factors can contribute to your vehicle deviating from the trajectory.

Also, make sure you have no SAS or ASAS parts on the craft, they are redundant with MJ and can cause it to get wonky.

Now, this is something new. Never once did those parts worked wonky with mechjeb, for me that is, nor were they redundant - I spend a lot of time with mechjeb off, relying on sas/asas when I want to.

[Capt'n Skunky]

Well, if you turn mechjeb off, you\'re not using it are you? I happen to like using MJ for everything so for me, having those extra control items do little more than add weight to the craft.

But as you said, without seeing his design, we are really making suggestions blindly.

Cheers!

Capt\'n Skunky

KSP Community Manager

[mockingking]

I\'ve tried a thousand different craft, and there has so far been no obvious rule as to which work better.

I\'m using a wide variety of mod parts, but keping things to Vanilla + Mechjeb does not solve the problems.

Are you making sure to change your ascent path final alt to match your desired alt? You have to set that in both locations or weird things can happen.

Hold on. Two locations to specify my desired altitude in? Do you mean the 'Orbit altitude' and the 'Turn end altitude' settings?

[Capt'n Skunky]

Hold on. Two locations to specify my desired altitude in? Do you mean the 'Orbit altitude' and the 'Turn end altitude' settings?

Yes, I think that\'s what they are called. I found if they didn\'t match, I would get really weird orbits.

Cheers!

Capt\'n Skunky

KSP Community Manager

[mockingking]

So, using some of the advice here - monodirectional RCS, no SAS or ASAS, setting the same altitude in both fields - I ended up with an apoapsis of 103km and a periapsis of 23km. The target was 71km. The rocket had plenty of power and fuel, no loose stages, and any parts that might inferere removed.

It was a manned satellite with Z02 tanks, Solar Power Generators and a small half-size fuel tank + quarter-size thruster, carried by seven full-size liquid fuel thrusters under 21 full-size liquid fuel tanks; six of the seven columns fed into the seventh and were discarded once the remaining fuel was down to three tanks.

There were no observable abnormalities, but somehow the Gravity Turn phase keeps on ending up so very wrong.

So, for the sake of experimentation, I just took one of those heavy jet cockpits, slapped a MechJeb unit, three tanks and a thruster onto it, put some SRBs below the whole cigar, and sent that thing up.

75km was the target. The resulting orbit was 74.8km on average, with 500m variations on apoapsis and periapsis. Inclination? Around 0,11 degrees. All of this prompted me to let loose a rather unfathomable curse.

So what is it that makes the first rocket go back down like a homesick dog, while the little cigarillo (despite having its boosters explode from overheating) does the most accurate orbit I\'ve ever seen?

PS: The smaller rocket even managed to do a perfect landing on its thruster without falling over. And there\'s 3/4 of the last tank left, too!

[Capt'n Skunky]

I suspect one of the plugins for those other parts is causing an issue with MechJeb.

What DLLs do you have in your Plugins folder?

Cheers!

Capt\'n Skunky

KSP Community Manager

[BlackBicycle]

75km was the target. The resulting orbit was 74.8km on average, with 500m variations on apoapsis and periapsis. Inclination? Around 0,11 degrees.

Don\'t you think it is a good result? Now a couple of well-measured farts with your RCS and you\'ll be on a perfect orbit.

[semininja]

What I typically use in the ascent AP is a turn end at the edge of the atmosphere and an angle of +5 degrees; that way, it\'ll coast up without trying to cancel vertical velocity first. I\'ve gotten consistent results at a wide range of orbit heights.

[mockingking]

For some reason, I now do have a heavy craft not unlike the manned satellite I described before in a good orbit. 500m variations, which is very much within tolerances...but why did it work now, but not before? The parts used were the same!

What DLLs do you have in your Plugins folder?

cart.dll, hstwLaser.dll, MuMechLib.dll, OrbitalConstruction.dll, Romfarer.dll, wheel.dll, Zoxygene.dll.

Don\'t you think it is a good result? Now a couple of well-measured farts with your RCS and you\'ll be on a perfect orbit.

I think that\'s an excellent result! I wouldn\'t even screw around with the RCS and risk overdoing it; it\'s quite good enough as it is.

Problem is, the craft I used to achiveve this orbit was completely useless for any actual work.

What I typically use in the ascent AP is a turn end at the edge of the atmosphere and an angle of +5 degrees; that way, it\'ll coast up without trying to cancel vertical velocity first. I\'ve gotten consistent results at a wide range of orbit heights.

Wouldn\'t that make the rocket overshoot its intended orbit?

Well, I\'ll try it right away. There\'s a Cluster ICBM I\'ve been itching to try, and so far it always ended up with wonky orbits. Maybe this will allow it to work!

Edit: I stand corrected. 0.11 degree of inclination, and around 40m variations on altitude. Looks like a 5 degree angle is an excellent idea!

Edit2: Which crafts do and which do not fly accurately seems largely unpredictable at the moment. Light (<50t) with no RCS, no Wings and no SAS/ASAS seem to perform best, while heavy vehicles with all the bells and whistles generally end up with high eccentricity and a death wish. Will try a light rocket with all the equipment next.

Edit3: Medium-Light craft with extensive manouvering equipment failed completely. Next experiment: Very heavy but primitive rocket.

Edit4: Yep, an equipment-less 160 ton mountain of a rocket managed to get a bloody good orbit going.

Edit5: It isn\'t winglets either. My suspicions are proven wrong! I now assume that it was insufficient thrust that doomed my vessels. From now on, everything gets enough thrust strapped to it to move the moon.

[mockingking]

Alright, this is the weirdest thing.

The same 160 ton rocket that managed to get a good orbit going (although rather slowly so) went completely apeshit on me once I added Fuel Ducts. I had previously observed that, for some reason unknown to me, the six outside colums retained their fuel reserves despite the seven internal ones being empty around the time the atmosphere was left. So, not wanting to waste seven thrusters, I attached the Fuel Ducts.

The rocket then proceeded to nose-dive into the KSC.

Explanations, please

[semininja]

I have seen some strange things happen with the ascent autopilot sometimes; I have a perfectly stable SE rocket that every once in a while goes into a flat spin and flings itself apart. I don\'t know what causes it, but sometimes stuff just conspires against certain rockets.

[Bluejayek]

I have a solution. This is a game about flying rockets. Fly them yourself manually

[mockingking]

I have a solution. This is a game about flying rockets. Fly them yourself manually

Tried it extensively, didn\'t really work. Well, it worked as long as I was happy with putting things in orbits that ended up looking like an unravelled ball of string! But apparently I\'m better off designing rockets and letting the MECHANICAL JEBEDIAH KERMAN take care of the piloting. I\'m just not very good at precision ???

Unless we\'re talking planes. With planes, I only ever automate when I need to let the thing fly straight across an ocean for a long time.

[vexx32]

... orbits never.... EVER... look like an unraveled ball of string. They always look pretty clean. It\'s the nature of physics.

That aside... I myself can achieve any form of orbit I wish to. I find the navball gives sufficient information to do just about anything you need to. The Map view provides any extra information that it does not.

The only thing that I find remotely difficult is rendezvousing, but it is possible. Very difficult, but by no means impossible. It is the only thing I would ever bother to consider using an autopilot for; the rest is relatively simple.

If you find it difficult, then all you need is a bit of practice; I don\'t even have a joystick.

[Endeavour]

I think you could use a little practice Mockingking

Really, the only thing I use Mechjeb for is the information it gives me because its annoying switching back and forth from the map screen. It tends to fail when you push 'auto do-something'

[mockingking]

But actually doing real precision work takes a lot of time and effort and trial and error. Also, it feels like 'work'

Seriously. Setting up Sattelite Networks, that I can do. But getting cargo shipped to an orbital shipyard? Timing launch to rendezvous with an orbiting craft, minimum accuracy of 10km? Repeatedly?

That\'s not just work, but hard work! ???

PS: I do have a policy of 'don\'t let MJ do it unless you did it yourself at least once'; but Rendezvousing (what a word!) is something that was just too frustrating.

[vexx32]

Maybe you oughtta look at some of the guides on rendezvousing.... they seem pretty well drawn up.

[mockingking]

God damn everything to hell.

So now I have a rocket that will get a wonderful orbit going, but only if launched at a heading of 270°. Launched at 90°, it confuses 'circularise' with 'ooops, forgot to close the windows, better head home'.

Wat the feces? >

[Markus Reese]

Personally, I manually set up near perfect orbits. How? I think I better make a video for that. I will post it up here after I make it, but to show you I can do it, this was something I showed off previously without any fancy mods. It was done with mod parts, but nothing that changed game mechanics. All visual and engines only.

The orbits are so accurate and circular, that the AO and PO are flashing all over the place and is impossible to actually read them. Also exact altitude that I want. 250,000m equatorial and 750km polars near exact, +/- 5 to 10 m pretty much.

[mockingking]

Sorcery! Burn the witch!