0.21.1 ASAS too weak.

[vzq]

When flying a massive craft with not-overwhelming amount of SAS strength, like 1st stage with non-gimbaling engines, it takes too long for ASAS to kick in with sufficient force to keep the craft veering off target.

From what i've gathered ASAS will use force in proportion to how much the course deviates from the desired one but when that little force is so low, that it cannot overcome the force causing the spin, it can easily turn 10 degrees before ASAS finally starts using enough force to bring it back.

When this happens in the middle of a burn to orbit it causes the final orbit to be several degrees off.

If i wish to achieve nice equatorial orbit i have to manually adjust to the deviations ASAS "ignores".

It's not all bad, of course i have noticed the great improvement in aircraft control and conserving RCS but in my opinion the ability to trust ASAS to keep the direction it was told to was worth these slight drawbacks, as that is it's primary role.

[NeilC]

There are two possibilities here:

1) this is a design problem, in that your craft has insufficient control authority. But you say that it will eventually correct, so....

2) Your design is out of the range that the ASAS controller is tuned for, and it's not dealing properly with your large ratio of inertia to control authority.

I would suggest a dev fix for 2) by either:

- allowing you to set ASAS gain manually with a slider during flight (larger gain = bigger response, faster motion, but more ringing and overshoot like the 0.20 ASAS)

- using a heuristic to set ASAS gain automatically based on craft inertia and control authority at max deflection.

Those aren't mutually exclusive, either.

[BigFatStupidHead]

I would suggest a design fix, and add more reaction wheels.

[chaos_forge]

I've also noticed that the .21 SAS is weaker, and most of the time it doesn't matter, but there are some instances where your craft is drifting significantly off course and the SAS barely does anything to correct it. I think that the PID term which increases force as the time spent off course increases should be tweaked.

[Scoundrel]

Sadly my SAS doesn't even work. It's slightly worse than just slapping on the old avionics package, as it will drift like crazy and doesn't even remotely come close to behaving like C7's demonstration video, even when I use the same/similar craft. Honestly I can't tell if it does anything. I keep wishing that, as an interim solution, tapping F would set a course target or something for the SAS to lock on to until it functions the way it's supposed to.

[Deejay2000]

Add more reaction wheels. With a big ship, I will usually add a reaction wheel before each decoupler for the launch stages. I think that should help at minimum.

-Deejay

[OdinYggd]

Don't hesitate to use the reaction wheels abundantly. They're not that heavy, and only absorb tons of power when actively steering- which when being used to counteract a lit engine they often can draw the required power from the engine's alternator so as not to drain the battery.

In the old days before rockomax gear, a good rule of thumb was one SAS per engine ignited at any given stage. The result would be a craft that could rotate freely throughout its available capabilities without using RCS unless necessary to move in a hurry.

It is very likely that a similar guideline could be followed with the new reaction wheels.

[Scoundrel]

It's not a lack of reaction wheels or control surfaces; when I follow C7's demonstration video and literally recreate the crafts he uses, mine do not behave like his do in the video. It's like a very weak avionics package, only without the ability to trim.

[carazvan]

Heavy is a relative term ... if you're building a 100t rocket sure no problem .. if you're building a small stage or SSTO there's another 5% weight for no reason.

Yes more reaction wheels help ... but let me put it this way:

- if the forum changed tomorrow to where you could only write 10 words per post .. sure you could still do what you need with 10 posts ... but that's not how it's suppose to work

Don't hesitate to use the reaction wheels abundantly. They're not that heavy, and only absorb tons of power when actively steering- which when being used to counteract a lit engine they often can draw the required power from the engine's alternator so as not to drain the battery.

In the old days before rockomax gear, a good rule of thumb was one SAS per engine ignited at any given stage. The result would be a craft that could rotate freely throughout its available capabilities without using RCS unless necessary to move in a hurry.

It is very likely that a similar guideline could be followed with the new reaction wheels.

[giltirn]

People, please stop saying 'add moar reaction wheels'. If you have sufficient control authority to manually hold a course then you should have sufficient control authority for SAS to hold a course. 99% of the people with SAS problems (including myself) also state that they have more than enough reaction wheels/wings/RCS to manually prevent the attitude from drifting.

[El wonso]

I recently left my plane flying east, after 20mins it was going west. So yeah, new SAS drifts away.

[Matt56]

After 20 min was it on the other side of Kerbin?

[El wonso]

Wat? How did you figure that out?

What I meant: I locked SAS at HDG 90, after 20mins it was 270something.

[m4v]

heh actually in 0.21.1 reaction wheels got quite a buff in their torque values, too much if you ask me. But I know what you mean, SAS seems to like to go towards prograde instead of maintaining your heading, my only answer to that is that the devs didn't want the SAS to be the crutch ASAS used to be, so you actually have to fly your contraptions instead of using autopilot.

[OdinYggd]

Note that there is no longer a distiction between SAS and ASAS.

The current SAS is intended to be the same functionality as ASAS used to be, while reaction wheels provide SAS-like rotational torque without any provision to maintain a heading on its own.

Nearly all of the pods and probe bodies include an ASAS controller though, that way the toggle SAS hotkey makes it attempt to track a heading.

Biggest difference though is if the ASAS detects you trying to turn, it will and let you do so before re-engaging to track the new heading.

[WedgeJAntilles]

I've had the same issues, especially attempting to hold attitude during launches. SAS seems happy to allow you to drift quite a bit before correcting, and during launch that can often end with a rocket flipping upside-down.

People, please stop saying 'add moar reaction wheels'. If you have sufficient control authority to manually hold a course then you should have sufficient control authority for SAS to hold a course. 99% of the people with SAS problems (including myself) also state that they have more than enough reaction wheels/wings/RCS to manually prevent the attitude from drifting.

THIS. This isn't an issue of insufficient control authority, it's an issue that ASAS doesn't seem to be taking advantage of that control authority.

[regex]

The current SAS is intended to be the same functionality as ASAS used to be, while reaction wheels provide SAS-like rotational torque without any provision to maintain a heading on its own.

Wait, what? SAS has the same functionality that the old ASAS used to have? You might want to have a talk with sal_vager then, because according to him it doesn't. Nor does it apparently hold a heading (E: even with more than enough control authority).

I am really confused...

Edited July 29, 2013 by regex

[Event Horizon]

Just give the ASAS/Inline module like 30 thrust or something, and give the big reaction wheel thingy 40.. that way there's some distinction between the reaction wheel modules.

[Scoundrel]

The problem is that the SAS doesn't "lock" to a heading, and it doesn't do any trim to maintain prograde for that matter. It just lazily noses about wherever it feels like. In other words, it doesn't even try to maintain a heading. It does not behave like C7's demonstration video at all. It's like it's not activating or something. The only thing it does do when in orbit is not fire the RCS at 100% power, though it does use the RCS to keep it stable, but it still won't pick a heading nor will it counter roll movement like it used to.

[regex]

http://forum.kerbalspaceprogram.com/entry.php/740-Updated-Information-on-SAS-in-0-21-1

Apparently those of us having these issues have computers with enough noise to never reach "locking" mode.

[chaos_forge]

How can I have noise in my input when using a keyboard . . . ?

[regex]

How can I have noise in my input when using a keyboard . . . ?

I couldn't tell you. Phantom noise from an old controller's drivers, faulty trackpad drivers on a laptop, who knows? C7 is looking into eliminating non-user generated noise though, which is a good thing.

E: I may be giving out bad information here, but the blog indicates that "noise" is what prevents locking mode from actually engaging.

Edited July 29, 2013 by regex

[littlebattler]

Yeah, this is a problem. I can manually control my old crafts and keep them stable just fine as if they were using the old avionics package, so that means the craft has sufficient control authority. Yet they still drift off course. The problem is that the new SAS barely deflects those control surfaces at all, and god knows how much force they actually eke out of those reaction wheels.

[regex]

Yeah, this is a problem. I can manually control my old crafts and keep them stable just fine as if they were using the old avionics package, so that means the craft has sufficient control authority. Yet they still drift off course. The problem is that the new SAS barely deflects those control surfaces at all, and god knows how much force they actually eke out of those reaction wheels.

According to C7's blog, an SAS in locking mode will use full control authority, if needed, to correct back to a heading. If your SAS is only trimming out movement (or acting "weakly") then it is in "damping" mode.

[Event Horizon]

The only time I ever turn on my SAS is when I'm already in the desired heading. Otherwise, I manually turn my ship to the heading, and turn SAS off when adjusting heading (I keep it on for smaller, more nimble crafts).

It seems to help with getting into locking mode, as well as giving ME control over the ship instead of an algorithm that hasn't had its kinks worked out yet.

Maybe a solution one day will be configurable SAS units, where you can test the ship yourself and edit parameters realtime or in the builder, and tailor the SAS to the ship. I believe that's how NASA does it with theirs.

Edited July 30, 2013 by Event Horizon