Autopilot unable to hold orientation toward special vectors

[Ed Gauss]

KSP Version: 0.90.0.0 32 and 64 bit

Platform: Arch Linux

What Happens: When commanding a ship to point at special vectors (progade, maneuver, etc.) using autopilot functions, the ship stutters and drifts off the desired orientation.

Steps to reproduce:

1. Enter "Impending Impact" Scenario
   
2. Ensure SAS is on, set to Stability Assist Mode
   
3. Command autopilot to orientate craft toward a vector, such as prograde
   

Workaround: This seems to affect crafts where reactions wheels have excessive influence over the vessel. 5kN of torque will keep a 2.5 ton vessel pointed prograde, but 1 ton vessel with the same amount will be pointed ~10-15 degrees off. Stability control still works without issue.

Other Details/Observations:

• This looks and feels like SAS behavior under FAR with high speed aircraft. I wouldn't be surprised if it's the same issue.
  
• All three Axes of movement stutter when this issue manifests. Stuttering occurs with mass/torque ratios above that required to suppress the issue, but is still proportional to the mass/torque ratio. The Kerbal-X stock craft does not suffer this issue until it looses all of it's propulsion stages, for example.
  
• Using RCS does not negate the issue. If anything, it makes it worse by reduce mass though propellant use.
  
• Probes and pilots both suffer from this issue.
  

Log File (from last session): http://pastebin.com/kcnRnvnf

[Claw]

This sounds like the old "SAS overcontrol" problem. A picture and craft file would help confirm that.

[Ed Gauss]

Will this do? http://gfycat.com/IndolentThinFerret

[Claw]

Actually that does help. I thought this bug had been squashed, but I will have to go back and check.

Thanks for the video.

Cheers,

-Claw

[lukeoftheaura]

I don't have this problem on ubuntu, so I can't help. sorry

[MOARdV]

I've seen this same bug on a Windows install (Win7 x64 / 32 bit KSP).

[Ed Gauss]

I made a comment on bug #3861, since this is probably an extreme case of that bug. I hope I didn't screw anything up.

[sal_vager]

SAS needs tweaking, and too much reaction wheel power just exasperates this issue.

All I can suggest is we try Regex's PID plugin.

[Ed Gauss]

More videos, showing how the torque-to-mass ratio of the vessel affects the issue: http://gfycat.com/BareMaleJanenschia http://gfycat.com/PeskyWeeDuckbillcat

Hopefully that demonstrates why I think this is an extreme case of bug #3861. Relevant thread: http://forum.kerbalspaceprogram.com/threads/103461-Power-and-RCS-are-drained-very-fast-when-using-Pilot-controls