'Hold this angle at all costs' SAS mode

[HalcyonSon]

20 hours ago, Vectura said:

The problem may be with low TWR. The thing is, if I have the rockets on max thrust they go supersonic very low in the atmosphere, then they get re-entry flames, causing me to lose control (usually using probes, and playing in 1.2). That aside, when going too fast I tend to lose some control authority.

General guidance is to aim for TWR = 1.5 at launch.  A few tenths higher or lower are okay depending on the size, weight, and shape of your rocket.  I usually hit 50 m/s around 2000 m for a 5 - 10 degree tip over.  I just launched a comet rendezvous mission with three Mansails that I built to go single stage to orbit and return the booster on parachutes.  I had to turn the gimbal down to 50 on all three to prevent oscillation.

[Vectura]

Ok, this particular rocket does not demonstrate the creep issue, but does demonstrate the issue of the craft dipping down when any control is so much as looked at. I mean, why does tapping a key to make it yaw make it pitch down several degrees?

Note: the 3 probe core payload is just a dummy for example purposes.

 

Edit: this craft seems to be an SSTO! Must experiment more...

Edited October 4, 2016 by Vectura

[Leafbaron]

2 hours ago, Vectura said:

Ok, this particular rocket does not demonstrate the creep issue, but does demonstrate the issue of the craft dipping down when any control is so much as looked at. I mean, why does tapping a key to make it yaw make it pitch down several degrees?

Note: the 3 probe core payload is just a dummy for example purposes.

 

Edit: this craft seems to be an SSTO! Must experiment more...

You can try turing off roll on your control surfaces, generally rockets only need pitch and yaw. Second you can try turning down the deflection of the control surface.

 

-leafy

[Stoney3K]

Those AV-R8 winglets have way too much control authority to begin with. You can turn the control surfaces of those winglets off and control the rocket with the gimbals of your Mainsail only.

SAS sometimes has trouble determining which control surface acts in which direction, that's why it frequently goes off course on planes. Either turn the control surfaces off or dedicate them to one axis of control only.

[problemecium]

This sounds a lot like the old SAS behavior, and I do want that back as an optional mode. In fact I'd like two new modes:

- Hold the current bearing as tightly as possible; if the ship is forced away from it, turn back as soon as it's possible to do so again.
This would help a lot with things like asteroids where the centers of mass and thrust are out of alignment and the ship starts to swerve during long burns. Rather than having to manually reset the heading, it'd be nice to have the SAS do it.

- Hold the current bearing relative to the parent body / navball; if the pilot turns to face, for example, heading 225 degrees at 30 degrees above the horizon, maintain heading at 225 and 30 even as the craft moves around in its orbit.
This would be very helpful in reentries and aerobraking maneuvers, in which the ship is likely to need to hold a specific angle relative to its descent path even as that path curves around with the terrain.

Alternately the old SAS behavior could be toggleable for all available SAS modes.

[StahnAileron]

9 hours ago, Vectura said:

Ok, this particular rocket does not demonstrate the creep issue, but does demonstrate the issue of the craft dipping down when any control is so much as looked at. I mean, why does tapping a key to make it yaw make it pitch down several degrees?

Remember that many engines have gimbaling and are affected by SAS control. If you try to yaw, the engine gimbals to yaw the vessel. For all you know, its gimbal angle was what was keeping you on course, not the fins. So if the gimbaling was what was holding your AoA (it's at least doing part of that unless you tunred off gimballing), you'd lose some AoA/attitude by pitching down. This is blatantly obvious with high gimbal range engines. (Like the VECTOR - My recent attempts to launch shuttle-like designs with it can fail in the dumbest ways because its gimballing is so extreme, both in range and responsiveness.)

Try turning off engine gimballing and see if the pitch problem is more or less severe. (Or limit which axis it can gimbal along. Pretty sure that's a thing for engines now...)

[HalcyonSon]

20 hours ago, Stoney3K said:

Those AV-R8 winglets have way too much control authority to begin with. You can turn the control surfaces of those winglets off and control the rocket with the gimbals of your Mainsail only.

SAS sometimes has trouble determining which control surface acts in which direction, that's why it frequently goes off course on planes. Either turn the control surfaces off or dedicate them to one axis of control only.

Ah yes, I forgot about that.  On aircraft, it's very important to decouple the control axes.  Elevators should control pitch only, Ailerons roll only, Rudders yaw only.  They'll get crossed up and give bizarre control inputs and tons of drag otherwise.  The same can happen to a rocket - if you're trying to roll and pitch at the same time, those are two different directions of engine gimbal.

20 hours ago, parameciumkid said:

This sounds a lot like the old SAS behavior, and I do want that back as an optional mode. In fact I'd like two new modes:

- Hold the current bearing as tightly as possible; if the ship is forced away from it, turn back as soon as it's possible to do so again.
This would help a lot with things like asteroids where the centers of mass and thrust are out of alignment and the ship starts to swerve during long burns. Rather than having to manually reset the heading, it'd be nice to have the SAS do it.

- Hold the current bearing relative to the parent body / navball; if the pilot turns to face, for example, heading 225 degrees at 30 degrees above the horizon, maintain heading at 225 and 30 even as the craft moves around in its orbit.
This would be very helpful in reentries and aerobraking maneuvers, in which the ship is likely to need to hold a specific angle relative to its descent path even as that path curves around with the terrain.

Alternately the old SAS behavior could be toggleable for all available SAS modes.

I think there's a mod for most of that... Persistent Rotation maybe?  I always thought it was bizarre that some spacecraft in this game will be facing prograde on one side of an orbit, and retrograde on the other side.  That just doesn't seem right.

[Red Iron Crown]

2 minutes ago, HalcyonSon said:

I always thought it was bizarre that some spacecraft in this game will be facing prograde on one side of an orbit, and retrograde on the other side.  That just doesn't seem right.

That's accurate behavior for a non-rotating craft in space, actually.

[HalcyonSon]

2 minutes ago, Red Iron Crown said:

That's accurate behavior for a non-rotating craft in space, actually.

Hmm, so that means every satellite pointed at the Earth has some initial rotation to keep it pointed "down."  Simple enough in reality I suppose, given that all the other orbital parameters are known.  Impossible in stock KSP from what I understand.  Time warp and re-loading killing all rotation... plus SAS wanting to kill rotation whenever it's on.

Fixing persistent rotation and preventing antennae from transmitting / receiving through satellites would make things a lot more interesting with SCANsat and Remote Tech.

[problemecium]

Yes, indeed. If a spacecraft does not rotate at all, then on opposite sides of its orbit it will be facing opposite cardinal directions (i.e. the opposite way on the navball). Real satellites that need to maintain orientation relative to Earth have a few different ways of doing this.
One such way is gravity gradient stabilization. One end of the satellite is made to weight more than the other, and because Earth's gravity is ever so slightly stronger on that end, a tiny tidal force is generated that keeps the satellite pointed the right way. This is, technically, simulated in KSP, but not applied during timewarp, so the only way to leverage it would be to leave the satellite in question focused at 1x time speed forever... but we can already guess that's a bad idea.