'Hold this angle at all costs' SAS mode

[Vectura]

Somethin I would like to see would be an SAS mode where the SAS tries its hardest to stay pointed at the angle it was facing in when enabled. Even if the player provides input, on release the craft would try to re-orient back to its initial angle. To actually change the craft's angle you would need to change the SAS mode or switch off the SAS, which can easily be done quickly with the F button.

I think this would be useful for those crafts where you want to stay pointing in one direction, but the SAS slowly creeps away and you have to keep correcting it, leading to a loss of efficiency.

[Corona688]

Stability assist does not slowly lose bearing -- what actually moves is you, relative to the planet.  it can be pushed off course to the point it stops at a new point, but it takes a large angular error to do so.

And yes, it would be very nice if stability assist took rotation properly into account but only in surface mode.  There's situations like docking where doing that would be very bad.

This may be harder than it sounds, if you don't want your craft going bananas when it overflies the pole.

Edited October 1, 2016 by Corona688

[flx]

Try Pilot Assistant

 

 

[StahnAileron]

I honestly think that in Surface Mode, the Hold Maneuver Node SAS mode should change to Hold AoA. What's the likelihood of someone wanting to do a maneuver in surface mode? (Surface mode usually means you're either on the ground or flying in atmosphere.)

I love Pilot Assistant; quite a powerful atmospheric auto-pilot, uh, assistant. Unfortunately, the UI for it needs help. It's not quite that intuitive and if you adjust the wrong thing, you can easily screw yourself over. (I've done this many times...) Problem is I have no idea how the UI for PA could be made easier to work with.

Still, KSP could use a stock Hold AoA SAS mode.

[Vectura]

Though I would like a mode for flying planes, I don't think that is the problem. The issue most likely lies in my rocket design, but I though this SAS mode could rectify it. The problem with a lot of my rockets is when doing my ascent, the attitude creeps down. If I so much as tap any rotation button the entire craft dips down up to 20 or even 30 degrees. I then have to hold a key to get it back to the prograde marker, but when I first hit that key it dips even further, then I waste even more fuel trying to correct. And when I get it back to the correct direction, there is a good chance the craft will just drop right back down again when I let go of the controls. I do not believe the problem is a lack of control because holding a rotational key causes the craft to fly around far faster than I would ever need. When this occurs in planes I can just angle the wings slightly to fix it, but for an inherently front-heavy thing like a rocket it is a problem.

Edited October 1, 2016 by Vectura

[EpicSpaceTroll139]

5 hours ago, Corona688 said:

Stability assist does not slowly lose bearing -- what actually moves is you, relative to the planet.  it can be pushed off course to the point it stops at a new point, but it takes a large angular error to do so.

And yes, it would be very nice if stability assist took rotation properly into account but only in surface mode.  There's situations like docking where doing that would be very bad.

This may be harder than it sounds, if you don't want your craft going bananas when it overflies the pole.

Actually, it in fact can lose bearing over time for reasons unrelated to moving around the planet. In a bit I'll get some screenie for you of a plane I have that I can hold straight on my own. When SAS handles it however, it waggles its tail up and down and slowly goes into a nosedive. This particular case is more of an instance of it constantly overcorrecting, but I've had other planes that would slowly change direction without waggling (like another one that would slowly roll). 

[Corona688]

3 minutes ago, EpicSpaceTroll139 said:

Actually, it in fact can lose bearing over time for reasons unrelated to moving around the planet. In a bit I'll get some screenie for you of a plane I have that I can hold straight on my own. When SAS handles it however, it waggles its tail up and down and slowly goes into a nosedive.

Ah, overshoot.  A 'hold this course at all costs' SAS would probably be even harder on the craft.  It would continue to oscillate well after the control channel saturated.

Quote

This particular case is more of an instance of it constantly overcorrecting, but I've had other planes that would slowly change direction without waggling (like another one that would slowly roll).

More specifics needed.

Edited October 1, 2016 by Corona688

[Superfluous J]

5 hours ago, StahnAileron said:

I honestly think that in Surface Mode, the Hold Maneuver Node SAS mode should change to Hold AoA. What's the likelihood of someone wanting to do a maneuver in surface mode? (Surface mode usually means you're either on the ground or flying in atmosphere.)

I don't like the idea of doubling the use of a button, but I support another button.

I could see holding a slowdown maneuver close to the surface of a planet (to come down to land) and midway through you realize it's holding your angle instead of the maneuver node. Probably not devastating but when you're moving fast near a planet's surface is NOT when you want your tools automatically switching what they do.

[StahnAileron]

Just now, 5thHorseman said:

I don't like the idea of doubling the use of a button, but I support another button.

I could see holding a slowdown maneuver close to the surface of a planet (to come down to land) and midway through you realize it's holding your angle instead of the maneuver node. Probably not devastating but when you're moving fast near a planet's surface is NOT when you want your tools automatically switching what they do.

My only question regarding that is at what altitudes on a vacuum body the system flips modes. Still, fair point on doubling up button use unless it a truly exclusive use case. I'd advocate both a Hold AoA (relative to prograde) and Hold Pitch (relative to the surface). These could be useful in orbit as well, I suppose, but having these two would keep the mode selection area consistent with even two columns. Slot them in just above the Target/Anti-target mode buttons to keep the consistent look since they'd always be available.

[EpicSpaceTroll139]

2 hours ago, Corona688 said:

Ah, overshoot.  A 'hold this course at all costs' SAS would probably be even harder on the craft.  It would continue to oscillate well after the control channel saturated.

More specifics needed.

I'll find my screenies and specifics as soon as I finish dinner. My point wasn't that this new SAS mode would fix it, but rather that not all slow drifts are due to motion relative to the planet; that the SAS does (sometimes) have problems keeping craft straight. 

Edit: Turns out I can't find a craft that the drift problem right now, but that's because whenever I encounter this I fiddle with the craft until I can get it to stop.

IIRC the last one I had that did the slow roll turned out to have its vertical stabilizer on slightly crooked. I find it understandable that this would make the plane want to roll. What I don't get is why the SAS let it roll even though just a touch of trim was enough to fix it when I flew manually.

Edited October 2, 2016 by EpicSpaceTroll139

[Corona688]

SAS responds to error input.  If it's not off-angle, it won't push.  So it can't do perfect correction.

[Stargate525]

Unbalanced thrust on crafts can cause this as well; never letting the craft drop into proper SAS, but keeping it in that weird dampening mode it has.

[Stoney3K]

The problem here is more the "Kill rotation" mode that SAS has (displayed by the two rotating arrows on the indicator light) which has a very nasty tendency to overcorrect or move your craft the wrong way.

I think this behavior is caused by the way SAS works in 'Stability Assist' mode, which pretty much holds the current craft attitude and tries to zero angular velocity at all costs.

Which is exactly what you don't want when you add a control input from the user. What happens is that the user deliberately rotates the craft to change attitude, which has little effect until SAS is saturated. The user then lets go of the stick, SAS senses an angular momentum and applies a control input the other way to try and kill that angular momentum -- which in turn causes overshoot and oscillations.

The way it should work is that user input should not be added to the SAS control output, but it should change the SAS attitude setpoint when the user applies a control input. When no input is applied, SAS should just keep attitude at the current craft attitude as it does now.

[Corona688]

Ah, I see now.

There's a big problem with moving the setpoint -- it'd be extremely easy to instruct SAS to put your craft on a trajectory it can't reach.

Perhaps if the setpoint just moved with the craft whenever QWEASD were released, instead of having to kill the saturation by tapping F?  It could even 'lead' it slightly.

Edited October 2, 2016 by Corona688

[Vanamonde]

This is what SAS used to do. I still miss it. There was some technical reason I don't remember why it had to be altered, but I'd still like the old mode as an option. 

[EpicSpaceTroll139]

@Vanamonde iirc it had something to do with the changes to SAS made to fix WR SSS (Wobbly Space Station Syndrome) but other than that I don't know.

[HalcyonSon]

On 10/1/2016 at 11:25 AM, Vectura said:

Though I would like a mode for flying planes, I don't think that is the problem. The issue most likely lies in my rocket design, but I though this SAS mode could rectify it. The problem with a lot of my rockets is when doing my ascent, the attitude creeps down. If I so much as tap any rotation button the entire craft dips down up to 20 or even 30 degrees. I then have to hold a key to get it back to the prograde marker, but when I first hit that key it dips even further, then I waste even more fuel trying to correct. And when I get it back to the correct direction, there is a good chance the craft will just drop right back down again when I let go of the controls. I do not believe the problem is a lack of control because holding a rotational key causes the craft to fly around far faster than I would ever need. When this occurs in planes I can just angle the wings slightly to fix it, but for an inherently front-heavy thing like a rocket it is a problem.

If your rocket is losing pitch with every control input, it probably means your TWR is too low.  By the same token, if it gains pitch with every control input, TWR is probably too high.  If it oscillates, either effect is worsened.  Lowering the amount of movement allowed on gimbaled engines or fins can help with oscillation.

[Vectura]

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.

[Leafbaron]

I agree with the SAS AoA hold, for spaceflight or surface mode. To me AoA hold sounds like the mech jeb feature where you can adjust the heading and the angle in that direction. So even as you are flying around the planet, you would hold at 0 degrees for example.

[Greenfire32]

On 10/1/2016 at 4:17 AM, Vectura said:

Somethin I would like to see would be an SAS mode where the SAS tries its hardest to stay pointed at the angle it was facing in when enabled.

That's exactly what the SAS does currently. What you're wanting is a more adaptive angle retention mode.

See, as you get further and further from the planet (and as the planet rotates underneath you), you get this effect that looks like the SAS is creeping away from where you set it.

The reality isn't that the SAS is creeping, it's that your position relative to the planet is significantly different. That's why you have to constantly adjust the SAS. KSP does have a "sort-of" solution to this in that you can set the SAS to follow your prograde marker but it doesn't do exactly what you're wanting.

What you're wanting is a new mode where if I set the SAS at a 45 degree angle relative to the surface, I want the craft to remain at a 45 degree angle relative to the surface. This would, in effect, require that the point of reference not merely be the planet, but rather the specific part of the planet directly below you.

Now I'm not sure if this is possible in KSP (I'm using SOI as my basis here), but I would think it should be. It would require the game to track your exact position over any given body, however, instead of just "at kerbin=true" so I would assume performance would take a small hit.

[Vectura]

The thing is, that is not the problem. Moving away from the planet does not explain the angle dropping a few degrees per second, ultimately culminating in a dive. If I turn off SAS and use trim it goes fine, though I have to keep a close eye on the roll.

[Red Iron Crown]

I have been a proponent of this for some time: When in surface mode the SAS should be in the rotational reference frame, not the inertial one. So under stability assist it would stay pointed at the same place on the navball rather than the same place against the starfield.

This and sun-relative SAS are the two biggest things I'd like to see for SAS now that the overcorrection jitter is largely a thing of the past.

[Corona688]

26 minutes ago, Vectura said:

The thing is, that is not the problem. Moving away from the planet does not explain the angle dropping a few degrees per second, ultimately culminating in a dive. If I turn off SAS and use trim it goes fine, though I have to keep a close eye on the roll.

Once again, it'd be nice to see the craft.  Right now I can't guess if this is a SAS related difficulty, a center of thrust problem, or something none of us are expecting.

[Vectura]

Can't get at my computer right now, but I will be able to towards the end of tomorrow. Here's to hoping that I remember!

[Leafbaron]

yes i agree with @Corona688, If your CoL is too far behind CoM and your control authority is not sufficient enough to keep the AoA, it's not necessarily that the SAS isn't working properly but that it just cannot get the command authority it requires from the control surfaces. Either increase control surfaces, or move CoL closer to CoM in addition if a roll is a problem you could try using an dihedral wing design. This design offers more stability but less maneuverability.