Spacestation weirdness

[LordFerret]

I'm wondering if anyone else has encountered this. I'm playing a Science game, v1.1.2, with a few mods (x-science, chatterer, freeeva, scansat, and kac) ... none of which I would suspect to be the cause of what's going on.

I have a spacestation I've assembled in orbit around the Mun (I'm very early on in the game mind you, without large docking ports unlocked yet). If I leave SAS on, the station begins to wobble after a few minutes, as though the SAS's on the various craft docked are competing and trying to tear the station apart. If I leave SAS off, the station begins to slowly spin, what appears to be on a random axis each time, if not on more than one axis. I can cancel the spin out by warping time, but the second I return to normal time the station begins (slowly) to spin again. Left alone, in no time at all the thing is really dancing around ... impossible to dock with. I have to quickly warp time a second to cancel out spin, jump back to real time, switch to the station and activate SAS, jump back to the craft I'm attempting to dock and hope I can do so quickly before the station starts trying to tear itself apart again - and then, yes, remember to turn SAS off again.

Edit: I forgot to mention, the nearly circular stable orbit has been shifting ... from an Ap/Pe @ 50km, to 49km - 71km, and I'm watching it shift about 1m on and off every few seconds.

Sound familiar? Some weird physics go on here.

Questions? Comments? Suggestions? Laughter?

Edited May 23, 2016 by LordFerret
forgot about orbit

[TrooperCooper]

 

Blame me for linking this old stuff, but while I think there are some fixes upcoming for stability in 1.1 sometime, I am pretty sure this super old bug never got fixed, since it goes down to the root of the engine: http://bugs.kerbalspaceprogram.com/issues/2002

Kasuha explained it to me this way:

 

Quote

What you are experiencing are most probably phantom forces, remnants of the so called "velocity Kraken" which used to tear apart ships that were moving too fast. Now, at or above 750 m/s absolute speed, a fix kicks in, making ships moving at that speed stable. However, ships that move slower than that are still moving using the old code which introduces random arithmetic errors into motion of individual parts. As a result, ships tend to rotate in random directions.

When you are in Mun orbit, you are moving at about 500 m/s - not fast enough for the "krakenbane" to kick in.

 

[egoego]

The shift of your orbit could be from the bug mentioned in this thread: 

Or be a result of your station spinning.

[TrooperCooper]

 

Yeah that orbit changing is something they are going to fix soon (tm) afaik

[qromodynmc]

I have similar problem, Not on stations but in my big rockets. If i use too many reaction wheel, ship starts to wobble. I dont think i had this problem in 1.05.

[Biff Space]

The change in orbit sounds like the phantom forces that other people have described, but the wobbling may not be.

If you are joining together large (high mass) units with the medium-sized docking port, you can get increasing wobbling, building up to proper thrashing about. (I don't know how bad this is now the switch to unity 5 has happened, I have modded the problem away, see below). It seemed worse if the parts are at right angles, so if you use a hub to connect two orange tanks at right angles with medium ports, it will wobble and start to thrash.

The solutions are:

1) short term, warp and de-warp. Like rotation, wobbling is stopped by non-physics warping.

2) long term, unlock bigger docking ports, and stick to small station modules until you have

3) (my favorite) install Kerbal Joint reinforcement, for sensibly rigid space stations.

HTH

[StrandedonEarth]

Maybe turn off some reaction wheels to reduce the forces SAS exerts on your station? I know I often need to reduce gimbal on my launchers so that prograde hold doesn't make things all wobbly.

Could also use KAS to strut your station to make it more rigid.

[Snark]

Several possibilities going on.

Trim.  This is a feature that I'd guess 90%+ of KSP players are unaware of, and wouldn't particularly need or want to use even if they were aware of it.  It's easy to activate accidentally; once you've activated it, stays on forever (for that ship) until canceled; and there's no UI way to tell when you've activated it, or whether it's currently active; and is subtle enough that you don't necessarily notice the problem until later.  What the feature does is to add a small amount of torque to the "zero point" of your ship-- i.e. it exerts a small torque continuously, even when your hands are off the controls.  It's possible you might have accidentally activated this feature without realizing it, which could explain the gradually increasing spin.  To cancel trim (i.e. reset it to zero), press alt+X.  This may or may not be affecting you, but it's worth a try.

(In case you're wondering "why on earth would anyone add such a useless, pointless, player-confusing 'feature' to a game?":  it's intended as an aid to flying planes.  If you have a plane that consistently wants to nose-down, for example, you can set trim to pitch up to compensate.  I've got no beef with that... it's just that the implementation leaves a lot to be desired, IMHO.  My personal preference would be to add an on-screen UI that pops up when it's active, with a visually obvious "center it to zero" control.  That way, even if a player has no clue what it is, they would at least know "hey, looks like I just activated something" and would have an obvious way to make it go away.  But I digress.)

About the increasing-thrashing-around-with-SAS problem:  There's a good chance that what you're experiencing is just an artifact of the way SAS works.  It has a sensor ("which way am I pointing?  is it the right way?"), and it has effectors (the "muscles" that it uses to try to make it point the way it wants).  The sensor, in this case, is whatever command pod, probe core, or docking port that you have designated as the "control from here" point.  The effectors are your reaction wheels.  If you have a big, floppy structure, and there are flopping docking ports in between the control-from-here point and the reaction wheels, then you have a problem:  there's a time delay in the control loop.  The probe core (or whatever) sees "hey, I'm pointing the wrong way" and sends a command to the reaction wheels, which start acting immediately... but because of the floppiness, there's a time delay before the torque from the reaction wheels changes the orientation of the probe core.  The probe core therefore keeps the torque applied to try to correct, longer than it should... which means that when it finally catches up to where it wants to be, there's too much spin in the system and then it tries to counter that, and again it overcorrects... basically you end up with a resonant system and hilarity ensues.

The fix for that problem:  Try to place your control-from-here point as centrally as possible on the station (i.e. not way out at the end of a sway-prone stack of modules).  Try to put your reaction wheels as close to the control-from-here point as possible.  Greatly reduce the overall amount of reaction torque in the system (it's a station, it's not supposed to move; just a little bit of torque to prevent "drift" is enough).  Use bigger docking ports-- the 1.25m ports are really not meant for space station assembly; the 2.5m ones are a lot stiffer.  And pay attention to the overall shape and mass distribution of your station; you can make a big difference based on where you put your mass relative to the CoM and to the docking-port joins.

My two cents, feel free to ignore here:  I would recommend against KJR as a solution.  It's not necessary; you can build big space stations just fine without it, if they're properly designed.  And my observation has been that using it often leads to other problems:  wobble is the game's way of telling you "maybe you need to rethink your design", and turning it on can just be a band-aid that leads to other problems such as poor rocket designs that have other issues.  That's not to say that KJR should never ever be used by anybody-- just that I think it tends to get over-used as a band-aid when it's not necessary.  People just look at wobble, and think "wobble bad!", and use it without realizing all the implications.

And then it's possible you could indeed be having some kraken issues, as folks here have suggested.  If you've got a station where you're not running any engines, and RCS is turned off, and your orbit is steadily and constantly growing (or shrinking) meaning that you have some thrust rather than torque happening... yeah, that's a kraken.

But I'd suggest checking the above issues, first.

[LordFerret]

Excellent feedback. Thank you, all.

The old bug reports, I remember discussion about those issues going back to like v0.25.0, although I didn't experience such things myself at that point, nor have I since up until now.

This is definitely a situation of things happening while not on rails, that much I've observed.

SAS placement and multiple unit contention is a thing I'm familiar with, and have had a lot of experience with in the past (wobbly rockets). I don't really believe this to be the issue, but it's a good point (again) to make note of and consider. Putting the 'control from here' point at the center of the station has no effect. Unless I'm missing something, I don't know of any way to pinpoint the station's center of mass, unless that's the point the view anchors to when the station becomes the active craft... that changes as crafts are docked or undocked.

The medium-sized docking ports are definitely an issue I was worried about from the start. I knew they weren't the ideal part to use in station building, but I don't have the large port unlocked yet ... it's a priority however. I will work with that goal before adding more mods (KAS / KJR).

The combination of poor-choice docking ports and SAS with attitude leaves my station looking like a wind-sock man flapping about in the breeze. Indeed, hilarity.

Trim! Now that is one I'd not thought of. I was aware of Trim being available, but have never intentionally used it... I use it in FS all the time however lol. I'm going to go check on that right now. It is highly possible I've fumble-fingered one of the Alt trim keys, I could very well have inadvertently activated that.

Edited May 23, 2016 by LordFerret

[benjee10]

I have exactly the same problem and it's incredibly frustrating as I just finished a long build of a gigantic interplanetary ship and now suddenly it is impossible to control, constantly rotating on a random axis. It's categorically not a design issue or trim being applied and the KER g force meter doesn't read any forces being applied, yet it rotates uncontrollably and is slowly going into an unstable orbit. A smaller ship I docked with the craft in an attempt to resolve the spin then also encountered the same bug; I was still able to keep it controlled, likely because the reaction wheel torque is relatively stronger on a smaller vessel, but its orbit began to change quite rapidly. 

[samstarman5]

I wonder if it might also be related to what happens to my craft during change in SOI. Such as when I go from Kerbin SOI to Mun SOI, my not-very-large craft will lurch and the Ap will suddenly have a significantly different number.

[Tourist]

When the Kraken swims by it creates ripples in the spacetime... not unlike how speedboat creates a wake which tosses a canoe around. By any chance do you keep hearing a alternating pattern of two bass notes? Survivors generally report hearing something like that.... those that survive anyway. Those that don't survive don't report anything... there is no time. 

Edited May 24, 2016 by Tourist

[LordFerret]

19 hours ago, benjee10 said:

I have exactly the same problem and it's incredibly frustrating as I just finished a long build of a gigantic interplanetary ship and now suddenly it is impossible to control, constantly rotating on a random axis. It's categorically not a design issue or trim being applied and the KER g force meter doesn't read any forces being applied, yet it rotates uncontrollably and is slowly going into an unstable orbit. A smaller ship I docked with the craft in an attempt to resolve the spin then also encountered the same bug; I was still able to keep it controlled, likely because the reaction wheel torque is relatively stronger on a smaller vessel, but its orbit began to change quite rapidly. 

Last night I deorbited the space station sections (all except for the lab) and began construction of a new one. This one will be a bit more symmetrical (in mass), and I'll see how that works out. I don't have high hopes for it though.

 

7 hours ago, samstarman5 said:

I wonder if it might also be related to what happens to my craft during change in SOI. Such as when I go from Kerbin SOI to Mun SOI, my not-very-large craft will lurch and the Ap will suddenly have a significantly different number.

That is something else I have noticed with the series of missions I've run out to the Mun. On return, I try to put my capsules Pe at 20k on Kerbin's back side. By the time I arrive, I find the orbital path well inside Kerbin, and then have to make a corrective burn (lest I come in too steep and cook a Kerbal).

[radonek]

21 hours ago, Snark said:

Trim.  …  What the feature does is to add a small amount of torque to the "zero point" of your ship-- i.e. it exerts a small torque continuously…

Whoa, hold it right there. Isn't trim supposed to change CoM? Such a hacky implementation would explain silence about it. Is it possible to disable it? I can't  find it in config file.

[Snark]

31 minutes ago, radonek said:

Whoa, hold it right there. Isn't trim supposed to change CoM? Such a hacky implementation would explain silence about it. Is it possible to disable it? I can't  find it in config file.

No.  Trim doesn't change CoM.  It just turns on your reaction torque a tiny bit.

I've been bitten by myself.  During experimentals, I had one particular space probe that wanted to keep tumbling.  Couldn't figure out what on earth it was.  Tried saving and reloading ... nope, persisted.  Since this was in experimentals, I just assumed it was a bug ("oh noes!  I have a phantom force!  kraken alert!") and reported it on the bug tracker, helpfully including a .sfs file for reference.  Couple hours later it comes back with a friendly "you have trim turned on, try hitting Alt+X."

Derp.

The hell of it is, I actually knew about trim, I'd run into it once long before.  It's just that since I never ever use the feature (and never will), and it had been so long, that I'd simply forgotten about it.  Trim is kinda sneaky, that way.

[pincushionman]

1 hour ago, radonek said:

Whoa, hold it right there. Isn't trim supposed to change CoM? Such a hacky implementation would explain silence about it. Is it possible to disable it? I can't  find it in config file.

This is a confusion in related terms - when  an aircraft is trim it simply means there is no net monent acting on it, which usually means the center of lift lift vector is alingned with the center of mass. Whereas trim is (in game terms, and also RL) whatever means of generating moment you have on hand to accomplish that - control surfaces, gimbal, reaction wheels, RCS - applied in addition to commanded control, such that you're trim when the contols are nominally centered.

Edited May 24, 2016 by pincushionman

[qromodynmc]

How do you open trim seems like nice feature.