(1.11) Trajectory changing while rotating ship

[adsuri]

While I was rotating my ship, my trajectory was changing , I am not using rcs thrusters and my craft isn’t complicated or flexible enough to bait the kraken. I have tried saving and loading, warping, and restarting ksp, which haven’t worked. It’s happening to other ships too and is making the game pretty hard to play

[VoidSquid]

Hi @adsuri and welcome to the forum!  

Your problem description is a bit vague, can you elaborate a bit more please? Maybe you have a screenshot/short video? And this issue is only with KSP 1.11.0, not, for example, 1.10.x?

[adsuri]

On 12/22/2020 at 8:44 PM, VoidSquid said:

Hi @adsuri and welcome to the forum!  

Your problem description is a bit vague, can you elaborate a bit more please? Maybe you have a screenshot/short video? And this issue is only with KSP 1.11.0, not, for example, 1.10.x?

Here: and this only happened a few days ago once I got 1.11

Edited May 23, 2021 by adsuri

[VoidSquid]

Hmm... reminds of an issue I once had with a space station. I somehow must have hit the wrong buttons and did set a trim, resulting in the vessel rotating slowly first, then faster and faster.  As soon as I entered time warp, the rotation was gone, only to start again the moment I disabled time warp.

If you're on PC, try ALT X, that resets any trim.

Can you load the save game and the vessel in KSP 1.10.1? Same issue there?

And there is a confirmed bug about a very weird behaviour of Kerbal doing EVA, see here https://bugs.kerbalspaceprogram.com/issues/26908 , maybe you found something related/similar.

Edited December 23, 2020 by VoidSquid

[Superfluous J]

I've noticed this too. It used to be just the way orbit lines worked and then they fixed it. I wonder if they removed that fix with all the work they did to make kerbals have mass inside capsules or something.

I don't remember version numbers, but IIRC the orbit line used to be calculated from the root part, not the vessel COM. So turning the vessel made the root part move, and the orbit was recalculating accordingly.

[adsuri]

Also to clarify, I am manually making it spin, it’s not like I’m stuck in some infinite rotation loop

[VoidSquid]

3 minutes ago, adsuri said:

I am manually making it spin

Oh... then I totally misunderstood you, beg your pardon.

[jimmymcgoochie]

If you put in a quick save (F5) and then time warp forwards to the Duna encounter with different orientations that produce different trajectories, do they actually follow those different trajectories or does everything converge on one specific trajectory regardless of what way the ship is pointing? Different things going on if it’s actually changing course or of it just looks like that in the patched conics but actually goes the same way.

I also find it interesting that the course deviations get bigger the faster the ship is spinning.

An image of the affected ship might offer some clues too.

[adsuri]

15 hours ago, jimmymcgoochie said:

If you put in a quick save (F5) and then time warp forwards to the Duna encounter with different orientations that produce different trajectories, do they actually follow those different trajectories or does everything converge on one specific trajectory regardless of what way the ship is pointing? Different things going on if it’s actually changing course or of it just looks like that in the patched conics but actually goes the same way.

I also find it interesting that the course deviations get bigger the faster the ship is spinning.

An image of the affected ship might offer some clues too.

When the rotation settles down it seems to always go back to the original trajectory, but any maneuver node to change my encounters don’t work bc when I tell SAS to hold the vector for the maneuver node the trajectory changes causing an infinite loop of the maneuver node vector changing and the game trying to hold at the maneuver. I’ll send a video of it when I can get on. Also here is the ship:

Edited December 24, 2020 by adsuri

[adsuri]

Something I would like to point out is that I did an unmanned Eeloo orbiter today and when I rotated the trajectory was only changing by a very little amount, even though I was very  far away where the trajectory is very sensitive to speed changes.

[18Watt]

On 12/22/2020 at 11:16 PM, Superfluous J said:

I've noticed this too. It used to be just the way orbit lines worked and then they fixed it. I wonder if they removed that fix with all the work they did to make kerbals have mass inside capsules or something.

I don't remember version numbers, but IIRC the orbit line used to be calculated from the root part, not the vessel COM. So turning the vessel made the root part move, and the orbit was recalculating accordingly.

I've noticed this in 1.11 also.   I also recall a much older version in which the effect was quite pronounced, it was extremely nice when that was fixed.  If it's back, it would be worth reporting.

I noticed it tonight while fine-tuning a burn to Jool, tweaking the arrival to get a Tylo intercept and a Laythe intercept.   So extremely small velocity changes were having noticeable effects that far away.  Like @adsuri, I was rotating, but not using any propellant (engines or RCS).   The effect I noticed in 1.11 is much less than the effect we used to get.   I can't remember how far back the old issue was either..

[paul_c]

I see this too, and just assume its due to the difference between two points on the vehicle (for example, centre of root part/controller and centre of mass makes sense) and when its rotated using the reaction wheel, these 2 parts have a relative velocity. When we're dealing with very slender intercepts which are set by very slender amounts one way or another, its going to show up as a change in the intercept values.

If you move a craft then "let it settle down again" they go stable again after a few seconds.

It seems fair enough and realistic enough - if you're aiming for such a slender intercept in real life, tiny things are going to upset it so a correction burn later on may well be needed to keep you on track.

[18Watt]

Not sure if this is the old behavior popping up again or not, but with the old bug (or maybe just 'unintended behavior') you could get free energy.   You would take a long ship with fuel tanks on each end, one full the other tank empty.   You would get the ship rotating in the pitch or yaw axis (not roll), and if you transferred fuel between tanks at the right time, you would change your velocity.   This was due to the way the game determined velocity, and I think it was eventually cleared up.  At least I haven't seen it in a while.

[vv3k70r]

Yes - this issue appear since 1.11 - new Kraken emerge.

[OHara]

I had the chance to look for this when I was intercepting Duna today in KSP 1.11.0

The only way I could see the effect @adsuri showed in the video, was to put a Kerbal on a ladder.  KSP does figure the forces between Kerbal and craft, and those forces show in the plotted trajectories --- at any instant the trajectories show where the bodies would go if the Kerbal were to let go at that instant. 

(The forces between Kerbal and craft are not quite equal and opposite in KSP 1.11.0 (bug report) but that effect is not the wobbling with rotation that @adsuri shows.)

From things I've done in the past, I am sure I would see the wobbling trajectories if there were loose cargo in a cargo bay, or anything else logically separate from a rotating craft but mechanically constrained to rotate with to it.   If anything like that is attached to your craft, @adsuri, boarding or docking it should make the wobble go away.

{Edit: I noticed the image of your craft was just a probe, so I built something similar and used cheat-menu rendezvous to put it alongside my craft on course for Duna.  When I rotate the probe, the trajectory wobbles just like yours, even though my craft can spin and have a stable trajectory.  Tomorrow I might try stripping parts on the probe until I see what was the key difference.

My guess is that this is a consequence of the "extra mass" bug (link) affecting several parts under 30kg. ]

 

On 12/25/2020 at 9:00 AM, 18Watt said:

I've noticed this in 1.11 also.

Was it a big effect, that we should get Squad to fix ?   Big enough to make a Duna intercept swing like the video linked by the OP ?   [Edit: Never mind.]

 

On 12/25/2020 at 6:28 PM, 18Watt said:

you could get free energy.   You would take a long ship with fuel tanks on each end, one full the other tank empty.

While I was rotating my craft I checked that you can still do this.  Rotate the craft and move fuel from a tank that is swinging retrograde, to a tank swinging prograde, and you have just increased your prograde momentum.  You can now show the same effect by moving cargo.   If we define 'bug' as code not working as intended, I guess this probably is not one, because I don't think the programmers even tried to account for the reaction forces on the imagined pipes that pump fuel.  This quirk almost seems more fun to have in the game.

Edited December 27, 2020 by OHara
tried copying the craft in the OP's image

[OHara]

I added a bug report (link)

In the thread reporting the related bug @Stamp20 pointed out a change with KSP 1.11.  Lightweight parts, under 30kg, are being rounded up to 30kg for some but not all calculations in the physics simulation. 

partRBMassMin = 0.03 // Minimum mass that a parts RigidBody can have - If this is too small then PhysX will not behave when it is dropped as a vessel, this is the default if minimumRBMass is not defined in the part cfg.

For several very lightweight parts, KSP places their mass at the parent part rather than the location of the small part itself,  so new players don't have to worry about mass balance with things like small batteries .   These parts join the parent RigidBody which is likely to be over 30kg so these parts are not affected by the recent change and cause no wobbly trajectory.

In the stock game, all parts under 5kg have this simplified physics, so one tempting solution is to change the rounding-up value from 30kg to 5kg.  Then only thermometers and other very lightweight parts---when dropped on the ground during  EVA construction so they have no parent and are their own RigidBody---get the extra mass.  Hopefully 5kg is enough.  It works for me so I put the change in a Module Manager patch on the bug tracker.