.22 SAS is now all WIbbely Wobbly (but not timey wimey)

[RayneCloud]

not sure what was wrong with the last asas.. seems to suck now, won't keep a heading like it did in .21

I have to entirely disagree with you. Rethink your designs.

[OdinYggd]

Can you share screenshots of the offending designs? I've found the 0.22 SAS to have a very nice response, smooth enough that I can actually cruise reliably without needing mechjeb to maintain proper heading. But SAS is useless without sufficient control authority to actually turn the vessel, and your design may require additional control systems to compensate.

When you finish adjusting your heading, tap F briefly so that the SAS toggles. That way it locks onto the position you wanted it at instead of trying to guess at it.

[Guest]

I have to entirely disagree with you. Rethink your designs.

It seems that many of the complaints stem from that fact that a previously working design gets borked when changes are made to sas by Squad. Which brings me back to the original question, why do they keep changing it? Sorry Damion, I'm not trying to troll you. Many here, including myself, respect your opinions.

Edited October 21, 2013 by Otis
clarification

[regex]

Personally, this really feels like complaining for the sake of complaining. People have complained about SAS for the last several updates, and C7 has dumped many dozens of hours into the system in an attempt to make it amazing. In my opinion, he's done just that. It's a hell of a lot smarter than it ever was, and a lot more capable. I'm not just saying this because I know C7, he's a good friend, that's true, but he's also extremely capable and has done a great deal of good with the stability assistance system. It's the best it's ever been.

I felt like complaints regarding the 0.21 SAS were definitely valid but I also see that version as an interim. I put in a bug report about it not holding a heading and C7 was kind enough to acknowledge the issue and fix it. I mean, sure, I want it to be a bit stiffer like 0.20, but I'm willing to accept the softer controls in favor of everything we gained in 0.21 (docking, launching, etc...) Now in 0.22 we have a very functional system that is adequate for pretty much every use. About the only thing that could make it better are tweaking controls for PID, and I don't see Squad adding that in.

[Payload]

Wow, to me, it's better than it ever has been. It is really very good now.

Perhaps some of you guys have multiple command pods or torque wheels placed about the craft. If you need more control, stack them in the same place. Don't distribute them around too much.

Use struts. The more rigid the frame of your rocket is the better it will behave. If you have a ton of gimbal engines try to lock the gimbal on most of them.

You generally only want the amount of force you need to get the job done with a bit extra for firm heading hold.

[Beachernaut]

I didn't really notice the SAS changes in .22 until I started building larger rockets. With the larger rockets it tended to "hunt" or "wobble" around where I set the heading. That said, it didn't take very long to learn the new SAS.

I found that if I do large changes an then hit F to lock it in, it would hunt. However, if I do large changes then use manual control to slow the changes as I approach where I want the heading to be before tapping F, it will stay on course. It was a simple modification to my playing style that I didn't feel warranted a post complaining about it. OTOH, we all play a bit differently, and have different expectations.

[DMagic]

It seems that many of the complaints stem from that fact that a previously working design gets borked when changes are made to sas by Squad. Which brings me back to the original question, why do they keep changing it? Sorry Damion, I'm not trying to troll you. Many here, including myself, respect your opinions.

Because there have been legitimate complaints about the way SAS works. The pre-0.21 definitely had problems and needed to go. But the 0.21 version had many issues of its own. The handling of unbalanced crafts is much improved in 0.22, and I think worth the change.

[JebidiahsBigSister]

Perhaps some of you guys have multiple command pods or torque wheels placed about the craft. If you need more control, stack them in the same place. Don't distribute them around too much

FYI - the placement of reaction wheels (or things with reaction wheels) does not matter with regard to SAS. They currently torque the same no matter where or how they are placed.

Also UPDATE to my original post - the "solution" seems to be moar reaction wheels which helps diminish the wibbelyness (my new made-up word for the day) and put things more in line with the old .21 behavior. As with much in KSP, Moar is better here as well.

Now will someone please tell me why the huge giant SAS thingy is lighter than the small one ???.

Edited October 21, 2013 by JebidiahsBigSister

[TNM]

i experienced it for the first time today. i undersized the "torque" by trying to optimize a small probe and i thought the integrated torque would be enough...nope. even in space, only having a slight asymetry in the probe (0.02 mass on one side difference), it couldnt hold the heading when activating the engine (less then 2t probe).

well, i will just oversize torque on my crafts in the future, should do the trick.

[Rickenbacker]

I did wobble a bit for me, until I figured out how it selects the new "zero" to hold. As soon as you release the steering keys, that's the spot it tries to return to, resulting in some wobbling. I gently tap the steering keys at the end point of the wobble, to cancel it out, and after that it holds the last position where I touched the key. Granted, I don't use MASSIVE rockets (not yet), but so far it works very well for me. Haven't even reinstalled MechJeb, as the in-game SAS does such a good job .

[RayneCloud]

It seems that many of the complaints stem from that fact that a previously working design gets borked when changes are made to sas by Squad. Which brings me back to the original question, why do they keep changing it? Sorry Damion, I'm not trying to troll you. Many here, including myself, respect your opinions.

C7 has been trying to build a system that accurately maintains control over a vessel using reaction systems and the vessels own control surfaces and control points, this includes gimbals on engines, RCS and more. For months now he's been trying to get to a point where the "computer" aka your SAS system, with enough reaction wheels, surfaces, or points of control can accurately maintain control over your vehicle without shaking it into oblivion. He's done just that in my opinion, and with any good game developer, you get "Iterative" development. Try something here, tweak something there, add something new, take it away if it doesn't fit or doesn't work.

[ChrisF0001]

It certainly seems to me that it's much more unstable than it was before. Sure, I can generally do what I need to do when I take direct control, but (not wishing to start a flamewar here) Mechjeb can no longer control the ship. It hunts, it wobbles through 90 degrees or more on either side of the target vector, it rolls the ship uncontrollably for no reason, it's just a disaster...

[Guest]

I happen to agree with you guys that say sas is working good now. I like it. I hope they keep it just like it is from now on. Once everyone adjusts their space programs to accommodate the new sas, we should be fine, and complaints and threads like this one will die a nice peaceful and painless death, allowing us to concentrate our efforts on other matters, like the hiccups.

edit : Thanks Damion! I for one am glad to see you around again.

Edited October 21, 2013 by Otis

[TNM]

i bring this up again because i noticed something today:

i oversized the torque on a probe of mine to get rid of wobble, but i noticed that, in orbit around mun, my probe drifted off course over time. that should not be possible? there is no drag, no engine burn, it just doesn't hold course. Maybe it is indeed not just wobbly, maybe there is a bug somewhere.

[Diazo]

i bring this up again because i noticed something today:

i oversized the torque on a probe of mine to get rid of wobble, but i noticed that, in orbit around mun, my probe drifted off course over time. that should not be possible? there is no drag, no engine burn, it just doesn't hold course. Maybe it is indeed not just wobbly, maybe there is a bug somewhere.

I'm assuming you mean over minutes.

That is a function of the co-ordinate system KSP uses.

SAS tries to hold you relative the to absolute co-ordinates 'box' the entire Kerbin system is in, not relative to you SOI.

So as you go around a planet in orbit, the navball will seem to drift as your orientation stays the same relative to the absolute co-ordiantes, but is moving in your SOI.

You should notice after an entire orbit that it has come back around and the navball reads the same again.

D.

[Jarin]

Sure, I can generally do what I need to do when I take direct control, but (not wishing to start a flamewar here) Mechjeb can no longer control the ship. It hunts, it wobbles through 90 degrees or more on either side of the target vector, it rolls the ship uncontrollably for no reason, it's just a disaster...

This just means that Mechjeb needs to update its control algorithms to account for the SAS changes. Such is the life of a mod developer on an Alpha-state game.

Personally, I think the SAS could stand to be tightened up a bit, but iteration is the name of the game, and I'm sure they'll continue tweaking.

[DMagic]

i bring this up again because i noticed something today:

i oversized the torque on a probe of mine to get rid of wobble, but i noticed that, in orbit around mun, my probe drifted off course over time. that should not be possible? there is no drag, no engine burn, it just doesn't hold course. Maybe it is indeed not just wobbly, maybe there is a bug somewhere.

What do you mean drifted off course? Like Diazo said, you stay oriented the same direction with respect to the entire Kerbol system, not just the planet/moon you're orbiting.

And actually, since the planets are also orbiting the sun, you will slowly drift even further off course. The easiest way to see this is to create a maneuver node, point at the blue vector and wait. After a few orbits you should notice that you aren't pointed directly at it anymore (the same thing happens with timewarp on). This is because the maneuver node is oriented with respect to the planet/moon's SOI, while you're direction is relative to the Kerbol system. If you try this again while in orbit around the sun you'll notice that you don't drift away from the blue vector anymore.

[Jean Deaux]

Seems to affect larger vessels. In 0.21 you would apply a bit of course correction, let go, it would more or less hold it. Now what I'm seeing is that I apply a bit of course correction, let go, and then it oscillates back *past* where I started. Utterly perverse!

The only way I've found to reliably counter it is to get the new heading, then apply opposite controls myself until the navball is stable, then press f to toggle SAS and lock in the new heading. Even then it still wanders a bit.

This is basically doing SAS' job for it, manually, and it's a bit fiddly.

Also noticing more monoprop being expended on docking because of this oscillation behaviour.

Same for me, get it centered and stable first seems to be the best performing option.

[Nostromo]

Not sure what kind of crazy crafts you guys are making but mine are very stable with SAS in .22.

[RayneCloud]

Here's the disconnect,

SAS has NO control authority of it's own, NONE, NADA, ZIP.

SAS is a "pilot" it moves the "stick" the control surfaces and control points respond to the stick being moved. If you have nothing on the ship that can gimbal, vector, or give any control authority at all? The ship wont move, wont stay on heading, wont anything and SAS can sit there and do things all day and nothing will happen.

SAS, for lack of a better way of putting it, pushes the WASD keys just like you do. Got a rocket with no control authority? You can press W all day long, and nothing will happen. Same thing with SAS.

[JebidiahsBigSister]

Here's the disconnect,

SAS has NO control authority of it's own, NONE, NADA, ZIP.

That's fantastic and thank you for explaining the difference between SAS and something like a reaction wheel but it has nothing to do with this thread in which a good many people have reported that SAS is behaving a bit wobbly - and it still is unless I "compensate" by putting a silly amount of SAS - opps I mean reaction wheels - on my larger ships.

[TNM]

tried to reproduce it, couldnt. will have to check my fraps, but if its not on there it doesn't matter anyway, a bug you can't reproduce isn't one.

Edited October 24, 2013 by TNM

[-RanZ-]

I have tried multiple builds and the new sas kind of sux... It went from really crappy pre-.21, to really good in .21 and back to somewhat crappy in .22. What was wrong with .21 sas that had to be changed?

[regex]

I have tried multiple builds and the new sas kind of sux... It went from really crappy pre-.21, to really good in .21 and back to somewhat crappy in .22. What was wrong with .21 sas that had to be changed?

It couldn't hold a heading. Even now it's having trouble with that, but it is much improved from 0.21. I've since tried larger builds and I find that it needs to be much stiffer for anything space-related, and it still doesn't use the full control authority given to hold that heading (I know this because I still have control authority left over to correct it). Basically it still feels like a trim system rather than something that holds my craft on course.