SAS Roll

[Daveroski]

Anyone else noticed a hyperactive roll on their ships?

It's like the SAS can't find a sweet spot or dead zone and goes too far and has to correct over and over.

My kerbals have all made claims for sea sickness and sold all their rocking chairs.

 

[moronwrocket]

Not really, no.  What does the placement of your SAS modules compare to the control point?  If they are far away, and your ship is like a giant noodle, the time for the twist to get to the control point could be long enough that it starts setting up a reinforcing beat.

[Demigod]

Roll? if you mean a rocking pitch and yaw as the rocket goes up yes. I think for me it has to do with engine gimbles as I'm using the kw rocketry engines. if I use standard it SEEMS fine,

Edited October 25, 2016 by Demigod

[Daveroski]

17 hours ago, moronwrocket said:

Not really, no.  What does the placement of your SAS modules compare to the control point?  If they are far away, and your ship is like a giant noodle, the time for the twist to get to the control point could be long enough that it starts setting up a reinforcing beat.

 

12 hours ago, Demigod said:

Roll? if you mean a rocking pitch and yaw as the rocket goes up yes. I think for me it has to do with engine gimbles as I'm using the kw rocketry engines. if I use standard it SEEMS fine,

Perhaps I should have clarified.

Length of play : Over somewhere between 4 and 5 thousand hours of playtime.

Level of play : I can dock pretty much anything to anything. RCS system preferred but optional.
I can land at a target (+- 100m) using the most fuel efficient methods.

I'm playing a new 1.2 vanilla career.

The SAS fault occurs on all vessels in space.

A small but noticeable twitch back and fourth on the ROLL axis of the vessel's control part. Regardless of part. Seems to happen less in a prograde or retrograde orientation but is still noticeable.

It affects even very small vessels with no SAS module other than the one in the command pod or core.

Pitch and Yaw seem to be ok.

It's as though there is no dead-zone for ROLL and it keeps trying to get to a zero state but keeps slightly overshooting and so goes back to try again. Making it overshoot by the same amount each time.

If you switch off SAS on a vessel the instant before docking it with a station or other vessel, the reaction is practically unnoticeable. Leaving it on when using larger vessels can cause a frightening oscillation that forces one to quickly turn off SAS and then manually turn off the reaction wheels on the newly attached vessel before turning SAS back on.

This is a problem affecting a vanilla (Never modded) installation of a new 1.2 career.

There you go.. clarified.

 

[Majorjim!]

Agreed I have noticed this too. 

[moronwrocket]

Daveroski - Thanks for the details, I'll keep an eye on it.  I haven't noticed this myself, but then again I don't do much rolling in space.  I usually only do it when docking, and having seen anything in particular. I've got a bunch more docking coming up and I'll see if it happens to me. Does entering/exiting timewarp make it go away?

[moronwrocket]

I installed my GOG KSP 1.2 for a clean setup with no mods, and increased the UI size to 150%.  I made a simple rocket with parachute, command pod, advanced in-line stabilizer SAS module, coupler, two FL-T800 tanks, a Swivel, and 4 AV-R8s for aerodynamic stability..

So, with all torque enabled, the roll/yaw/pitch return from railed back to SAS in HOLD mode takes about 6 seconds to damp out completely.  There might be some microtorque, but after 10s there isn't anything noticeable to my eye.

I disabled the advanced inline stabilizer, and it was still able to stabilize the rocket after 10s.

I reduced the pod's reaction wheel torque to 10, and then it does take quite a bit longer for rolls to dampen down, and interestingly, there is a strange behavior noticeable.  As the SAS engages and counters the rotation, it stays at a low value until the rotation is almost stopped, and then pops up to almost half, before switching direction.  Seems like there is an issue with the algorithm when it pops higher and then switches direction, probably when it is close to the hold position.  I put up a short video at Example of SAS behavior.

Looks like a legit problem, esp with low SAS authority.

One interesting thing I noticed, the indicators at rest don't point to the center of the middle hash mark, but to the left edge. I wonder what that means for the navball indicator.

(bottom edge for pitch)

 

[Daveroski]

20 hours ago, Majorjim! said:

Agreed I have noticed this too. 

Ah. Excellent.

13 hours ago, moronwrocket said:

Daveroski - Thanks for the details, I'll keep an eye on it.  I haven't noticed this myself, but then again I don't do much rolling in space.  I usually only do it when docking, and having seen anything in particular. I've got a bunch more docking coming up and I'll see if it happens to me. Does entering/exiting timewarp make it go away?

Only for the duration of the warp as one would expect.

6 hours ago, moronwrocket said:

I installed my GOG KSP 1.2 for a clean setup with no mods, and increased the UI size to 150%.  I made a simple rocket with parachute, command pod, advanced in-line stabilizer SAS module, coupler, two FL-T800 tanks, a Swivel, and 4 AV-R8s for aerodynamic stability..

So, with all torque enabled, the roll/yaw/pitch return from railed back to SAS in HOLD mode takes about 6 seconds to damp out completely.  There might be some microtorque, but after 10s there isn't anything noticeable to my eye.

I disabled the advanced inline stabilizer, and it was still able to stabilize the rocket after 10s.

I reduced the pod's reaction wheel torque to 10, and then it does take quite a bit longer for rolls to dampen down, and interestingly, there is a strange behavior noticeable.  As the SAS engages and counters the rotation, it stays at a low value until the rotation is almost stopped, and then pops up to almost half, before switching direction.  Seems like there is an issue with the algorithm when it pops higher and then switches direction, probably when it is close to the hold position.  I put up a short video at Example of SAS behavior.

Looks like a legit problem, esp with low SAS authority.

One interesting thing I noticed, the indicators at rest don't point to the center of the middle hash mark, but to the left edge. I wonder what that means for the navball indicator.

(bottom edge for pitch)

 

I'm happy that others are now as perturbed by this as I.
I tend to spend far too much time then trying to figure out what the problem is rather than keep my mindset in the game-world. Completely spoils the flow for me.

The strange thing is that it doesn't seem to be constant. Perhaps a particular part is responsible? Perhaps symmetry on radial attachments? I'll have to try and notice which parts are on the vessels and generally see if the worst offenders have something in common.

Edited October 27, 2016 by Daveroski