Looking for Wobble Fix - Coupled ships

[Jean Deaux]

I've been playing more with coupling ships for travel now that I've gotten space plane designs now and I'm finding the easiest solution to deliver my planes to distant worlds is to use another vehicle for towing to deliver and then return the space plane. In previous rocket designs I've done, I've avoided joining ships together because of the wobbling I get with ships coupled together, but now that I'm finding a need to have such designs, I'm hoping one of you may have a tip on how to minimize, if not eliminate, this wobble.

I've tried turning on and off the reaction wheels in various stages, but that doesn't seem to have any affect on controlling the wobble. Check out the captured video of a ship I'm testing, see if you can spot any designs or features that may help out. Location is in solar orbit and out of the SOI of any planet or body.

[Sirrobert]

SAS is not on, but there is reaction wheel input.

Are you giving that input? If not, that's the cause of the wobble.

Alternativly, is Mechjeb autopilot currently in control? That might also be the cause of the wobble.

While this should be fixed by simply stopping input, Mechjeb tends to overreact in situations like this. So it'll try to correct by oversteering in the wrong direction, amplifying the wobble

[Jean Deaux]

Yes, MJ was in control during the video as I tend to let MJ manage the long burns while I do other things, which is why the SAS wasn't lit up. I agree MJ exaggerates this issue, but even flying manually it still wobbles a good bit.

[Master Tao]

Sounds like MechJeb is exaggerating the wobbliness of the Clamp-o-Tron Jrs. Try shifting those ports far behind the CoM and using larger ports.

You can see something making input according to the indicators in the bottom left corner. Since it's not you and it's not SAS, it has to be MechJeb that's unable to dampen the wobbles.

[Daid]

Quantum struts could help you in this case, to make everything more rigid:

http://forum.kerbalspaceprogram.com/threads/71030-Quantum-Struts-Continued-(2014-07-24)-24

[biohazard15]

It's simple: your design is too bulky, and payload's CoM is too far from docking port. You should use 2.5m docking port, and you probably should put your payload in front (push it instead of pulling)

[Sirrobert]

It's simple: your design is too bulky, and payload's CoM is too far from docking port. You should use 2.5m docking port, and you probably should put your payload in front (push it instead of pulling)

If there's a difference at all, pulling is better to prevent wobbling than pushing

[Laie]

Mechjeb tends to induce wobble. Most of the time, it's not a problem and often barely noticeable. But it gets worse the more flexible your craft is, and can portentially become catastrophic (I once had a ship where one piece swang like a church bell, +-120 degrees. Miraculously it didn't come apart, but when it collided with other parts of the vessel, the situation deteriorated rapidly).

In this particular case, it wouldn't be fair to blame mechjeb -- this is an articulated bus, what did you think would happen? I suggest you try a larger docking port, or even multi-node docking. Or some mod like KAS that will let you place struts in space.

[Targa]

Quantum struts could help you in this case, to make everything more rigid:

http://forum.kerbalspaceprogram.com/threads/71030-Quantum-Struts-Continued-(2014-07-24)-24

THIS ^^^. Was annoyed for a long time about nasty wobble with landers on top of rockets. Quantum Struts will make everything solid as a rock. This is the one mod I would put on my "must have" list.

One caveat: Very, very rarely, if you have quantum struts activated on a ship and then dock with another ship, they may explode violently. Save often! This has only happened to me once in a couple of months of daily play, so it's not likely to happen much, if at all. There is also a known bug that if a quantum strut is aimed at a part with no physics, it can prevent your ship from being able to take off. See the above-linked forum thread for an explanation of what parts not to aim at.

This mod is still worth it's weight in gold.

[capi3101]

It's simple: your design is too bulky, and payload's CoM is too far from docking port. You should use 2.5m docking port, and you probably should put your payload in front (push it instead of pulling)

I would agree that OP should be using a larger docking port given the size of what he's hauling around. That said, pulling the payload as a rule produces less wobble than pushing in my experience.

Only other thing I can think of trying would be to bring that set of engines closer in to the centerline of the craft; the further out the engines are, the more unwanted torque that will be produced when your SAS attempts to compensate for the wobble.

Quantum struts might help here; do they stay attached when the payload undocks, or come back into play when the payload re-docks? I don't know enough about the mod to say one way or the other...

[Claw]

You might also want to try setting "control from here" to one of those smaller ports in the middle. Or at least move it around from where it's at (I'm guessing it's that top port). Turning off reaction wheels on the lander will also help, but it looks like you were doing that already.

Cheers,

-Claw

[Sirrobert]

I would agree that OP should be using a larger docking port given the size of what he's hauling around. That said, pulling the payload as a rule produces less wobble than pushing in my experience.

Only other thing I can think of trying would be to bring that set of engines closer in to the centerline of the craft; the further out the engines are, the more unwanted torque that will be produced when your SAS attempts to compensate for the wobble.

Quantum struts might help here; do they stay attached when the payload undocks, or come back into play when the payload re-docks? I don't know enough about the mod to say one way or the other...

You can turn them on and off whenever you want.

I do believe they also turn off when you undock

[Jean Deaux]

You might also want to try setting "control from here" to one of those smaller ports in the middle. Or at least move it around from where it's at (I'm guessing it's that top port). Turning off reaction wheels on the lander will also help, but it looks like you were doing that already.

Cheers,

-Claw

Yep, tried that too, with no improvement.

[Jean Deaux]

Quantum struts could help you in this case, to make everything more rigid:

http://forum.kerbalspaceprogram.com/threads/71030-Quantum-Struts-Continued-(2014-07-24)-24

This is my fallback option, I'm familiar with and have used Quantum struts in the past but I'm also trying to use as few mods as possible simply to avoid bug issues within the game that seem to pop up when running mods. I'll only add a mod for a necessary feature the core game lacks, and it's quickly looking like this may be one of those instances. I am hoping there are other options before I go the mod route that I can try.

[Sirrobert]

This is my fallback option, I'm familiar with and have used Quantum struts in the past but I'm also trying to use as few mods as possible simply to avoid bug issues within the game that seem to pop up when running mods. I'll only add a mod for a necessary feature the core game lacks, and it's quickly looking like this may be one of those instances. I am hoping there are other options before I go the mod route that I can try.

If you use KAS (which I concider a feature the core game misses), it also has strut ends that you can place and link yourself

[capi3101]

If you use KAS (which I concider a feature the core game misses), it also has strut ends that you can place and link yourself

I don't remember that; have they updated that mod in the last few months?

[Yasmy]

In addition to the fine comments above, I'm going to give the standard advice: moar struts.

Or, at least, better strutting.

Specifically, those side arms appear to only be restrained radially. They could use some axial (lateral) restraint to prevent side to side motion. It looks to me like the attached vehicle is doing a good job of indicating lack of rigidity in the main craft.

[Laie]

[Quantum Struts] is my fallback option, [...] also trying to use as few mods as possible

In that case, multi-node docking. Don't need to be many, three will do (think of a three-legged stool), but place them as far apart as you reasonably can. Each at the end of a girder or i-beam or something.

[Sirrobert]

I don't remember that; have they updated that mod in the last few months?

Not sure about updates, but it's here now

[merendel]

Well one design choice option that could have some effect would be shift the 2.5 reaction wheel down to the other end of the tank, they generaly are more effective and behave better the closer to COM they are. However it looks like something else is going on as your getting wobble before you even start thrusting. Make sure MJ is compleatly off, no SmartASS, no other utilities or tweeks active. heck disable every reaction wheel on the ship and do the timewarp trick and see if the wobble starts up agian.

One other thing you could try is enable just the reaction wheel on the lander (the advanced module) and controll from the docking port on top of it. Dissable all other reaction wheels and the LV-Ns gimbles. This would result in just the lander applying any torque to the ship, if it holds itself steady nothing else should be moveing.

Honestly though I've had alot more trouble with random SAS wobble in general since .24 came out. Even ships that are one solidly strutted together piece tend to oscillate a bit as I'm trying to follow a maneuver node. I can only imagine how bad it would be with a pendulum hanging off the back of my craft. I suspect MJ would make it even worse if I used it anymore. If nothing else works your best bet is probably going to be either useing larger docking ports for the rigidity and or some mod that allows strutting for docked ships (quantum struts or KAS would be my recommendations)

edit: Also one other thing that bothers me on a second look, why'd you design that ship with 2 sets of LVN's if your only going to use the outer set as drop tank first stages? You'd be better off with just the main 4 and a set of engienless drop tanks as you wouldnt be lugging around 4 extra engiens that are not fireing for that first transfer burn.

Edited August 12, 2014 by merendel

[sal_vager]

This trick isn't 100%, but it can help....

Use landing legs (LT-1's) to brace your craft, they can push against parts of your rocket and against each other, and of course they fold away when not in use

[Jean Deaux]

edit: Also one other thing that bothers me on a second look, why'd you design that ship with 2 sets of LVN's if your only going to use the outer set as drop tank first stages? You'd be better off with just the main 4 and a set of engienless drop tanks as you wouldnt be lugging around 4 extra engiens that are not fireing for that first transfer burn.

I've set-up the outer quad LVN's for the "to" trip and the inner quad LVN's are the "return" trip with the outer's discarded. I could add a fuel line from the outer to the inner to fire all the engines at once then discard the outers once dry, and is one of the design aspects I'm currently testing, to not have "dead" engines during a burn. By not having this fuel linkage, it gives me plenty of time to get to the coupling to disable crossfeed so that I'm not depleating the fuel from the "drag-along" lander during transit.

[merendel]

I've set-up the outer quad LVN's for the "to" trip and the inner quad LVN's are the "return" trip with the outer's discarded. I could add a fuel line from the outer to the inner to fire all the engines at once then discard the outers once dry, and is one of the design aspects I'm currently testing, to not have "dead" engines during a burn. By not having this fuel linkage, it gives me plenty of time to get to the coupling to disable crossfeed so that I'm not depleating the fuel from the "drag-along" lander during transit.

Ironicly it might actualy be helpful to let the lander drain of fuel and just refill it off the main stack before decoupling. it would be lighter and have less inertia to try and counter if wobble happend.

I've built tug ships of a somewhat similar design myself. 2 or more LV-N's radialy out from a 2.5m core with docking ports on either end of the core. However generaly what I do is stick the payload on the underside like you and just mount a 2.5m tank on the front docking port of apropreate size for the trip. the fuel lines feeding the drive take from the furthest tank first, in this case the front drop tank. When its empty I just decouple it and let it float away. tad more tricky if it needs to be decoupled halfway through a significant burn but normaly it can be planed for, cut drive for a few moments, decouple, possibly with a bit of rotation, get out from behidn the tank and resume the burn. This also has the advantage of makeing the tug reuseable for a later mission. once back in LKO just send up a pair of orange tanks, 1 to fill the tug and 1 to be a droptank for the next mission. Heck most likely I'd have several full orange tanks hanging off a fuel depot in LKO just waiting to be plucked off by a tug when needed.

I also favor shorter fuel tanks on the engien pods and a longer core. Big risk of your design is accidently bumping an engien while docking/undocking the payload. I dont like them extending back far enough to bump the payload.

[Jean Deaux]

For all those worried about Bill, you'll be happy to know he and his rickety rocket successfully completed their trip to Gilly and back. If you've yet to see aerobraking, I captured the moment so you can see the benefits of doing so.

[NoClass]

This trick isn't 100%, but it can help....

Use landing legs (LT-1's) to brace your craft, they can push against parts of your rocket and against each other, and of course they fold away when not in use

That's... interesting!

The only input I have is something I read (can't remember where) about how SAS torque is applied in the KSP engine. That is, SAS torque is applied directly to the center of mass of the craft, rather than on the actual torque part. So, SAS parts on the wobbly section wont stabilize it unless the CoM is in the wobbly section.

I'm curious about how SAS works WRT control surfaces, RCS, and gimbals on craft that can flex. Does SAS recognize when the control part has been moved relative to the CoM, and adjust input accordingly? Or do control parts move based on where they are in the craft file?