Come back old ASAS - all is forgiven!

[Nodrak]

Ok after much testing for me at least I think I have identified the 3 key issues that current 0.21 ASAS has:

Problems

1) After setting a course ASAS doesn't always hold it if it's not level flight -it seems like the ASAS is just a little too slow to react

2) Manually inputing direction to ASAS turn both other control directions to 0* - this is likely a bug since it was stated that new ASAS shouldn't interfere with the other axes you were not touching

3) After manual input to a direction in ASAS control moves it to 0 then slowly tries to move it back to a direction to counter - this is especially bad for trying to do gravity turns or pitch adjustments to precise locations

Solutions

1) After looking at the imputs I think a 2x adjustment to the input it provides should be enough to stop this from showing up and it shouldn't make for a huge change to fuel use

2) Do what was said where only the axis you touch get changed others keep under ASAS control

3) The 0.20 system isn't great here either to where the needle moves like crazy around the direction once you let go (wobble cause most likely) but maybe instead of starting with 0* as the default search for the ASAS start with the last position of the control input.

And an album to show off how problems can arise ... didn't mean for it to be night/day between 0.21 and 0.20 but it does work nicely ... look at the bottom left control surfaces input and at the bottom center for the navball:

If you are trying to compare .21 and .20 with those pictures, you should probably make sure they happen within the same magnitude of atmosphere... All this is showing is the obvious? Space plane in low atmo without RCS having slightly harder time controlling pitch in plane that wants to pitch down. AKA, you have less magic torque.

I just flew to the moon with Kerbal X to test, works like a charm even though Kerbal X is tiny off balance. Only issue is SAS is slow to respond to manual input, as you noticed. Nothing crippling, and actually works exactly how the OLD ASAS did, you just expected it because its this T button you hit...

[Sathurn]

First impression of the new asas, not a clean install but with stock parts. Using a small orbitor

I feels weird, I caught myself turn in on and off.

Trilateral symmetry seems to introduce a spin when you try to change just one axis. Easily canceled out.

ASAS does not stop rotation very well, but does prevent unwanted rotation from starting just fine. So when you turn to a heading you must cancel the rotation manually.

Using intentionally out of balance rcs, Heading change 1-2 degrees during translations burns. Returned to proper heading when burn ceased.

It feels weird to me, but I like it.

[UmbralRaptor]

I've been having various issues with the new SAS. However, I've found that the old ASAS still exists. As such, I've thrown together a part that brings it back.

This version is tuned differently from 0.20.2's to reduce overshoot and jitter, but I've included instructions if you dislike it.

Edited July 25, 2013 by UmbralRaptor

[AlternNocturn]

I'm building a miniature space station in orbit around the Mun, and not once have I had a problem with the new SAS regardless of the fact that I'm still getting used to it. I don't understand the complaints.

[Anton P. Nym]

So far I have few complaints; most of the designs I tested that had trouble were probes that relied on the pod having gyros. Since they don't anymore, that's expected behaviour. I'll write those probes off and start anew.

The only design that perplexes me is the RATO I used to boost a solar glider in 0.20 now induces a consistent slow yaw to port that will plant the glider firmly into the VAB's door unless I jettison the RATO early. Since the RATO had no guidance logic or control surfaces, I guess this shouldn't surprise me; I'll do a quick redesign with some fins and see if the problem persists.

-- Steve

[EdisonTrent]

https://www.dropbox.com/s/frj8dyqbvb2i0oh/_21%20SAS%20Test.craft

Here is an example rocket.

The first stage it looks like the 4 fins are just not cutting it. Probably working as intended.

The second stage on the other hand, with 4 gimbaling engines has plenty of controlability when flown manually. If you fling the rocket a certain direction you can easily slow down the movement and correct it by imputing controls in the opposite direction.

But the SAS can't. It sluggishly does do it eventually when movement has slowed to a certain point (while rolling slightly) but clearly not as easily as manual correction. Maybe the SAS is not using gimbaling enough and is relying on capsule torque.

Because the last stage works wonderfully and if it was always like that then I think everyone in this thread would be happy.

[SkyHook]

The new ASAS works flawlessly for me. I have found that any failure of the ASAS to hold heading and pitch are due to design errors or unsustainable flight parameters. As for those who want the old ASAS back...I simply fail to understand it. It is an inferior system and the New ASAS-bugs and all-is an improvement upon the old one.

[malkuth]

curious if this is a clipping issue with people using vessels with lots of clipping.. Not to sure.. i used HalsFury Gemini 2.. And it works pretty dam near perfect.. Some of my other vessels I noticed a drift though.. But I combat it with turning on SAS and back on like the old days. Seems to work.

[Oxydin]

This is a plane. It's got control surfaces and is a very stable aircraft. It also has reaction wheels because I pulled it over from 0.20 and its ASAS unit magically became reaction wheels.

It's doing something when I turn it on, but that something isn't doing the job of holding a heading.

From what I've heard, SQUAD have tried to facilitate earlier crafts, but have stated that some may not work. Maybe try a fresh build? I've honestly not tested 0.21 yet (I'm about to) so this is just speculation.

I do find the posts about 0.21 abit ironic though. We have all shouted at squad for the last 3 weeks to release 0.21 and now that it's releasesd, we're complaining about it.

[RocketPilot573]

Well this is weird, how do some people have working SASs and others not? I know, how about we all post pictures of the ships we are using to find out if there is a problem or not. Because I'm convinced that there really is something wrong.

[Benie]

From what I just read in a well-written tutorial (here), ASAS is gone. We've pretty much gone back in time before it even existed. It shouldn't even be called "ASAS" anymore.

Reaction Wheels are realistic, but this is supposed to be a Sandbox game where you can do what you want. We've all been used to ASAS doing the flying for us (in a sense). This is not a bug. I think this was an intended feature of what's to come in the future of this game. I just wish they could had explained better than "ASAS holds your heading nomatter what you do! Fly cheat mode for our community! Woot!", causing hype for the beloved change. Even I was fooled by it (what I said is how I feel, going by a video I saw somewhere in this forum).

I personally don't like hearing this when I saw this thread pop up, but I can't do anything about it. Whining and complaining to SQUAD isn't going to do any good. Yes, we could boycott, but what good would it do? As I said before, and I'll say again, this is their game.

[FleetAdmiralJ]

From what I just read in a well-written tutorial (here), ASAS is gone. We've pretty much gone back in time before it even existed. It shouldn't even be called "ASAS" anymore.

There are reaction wheels, which will give you torque, and then there is the "advanced stabilizer" which actually does the controlling for you like the ASAS did before. Or, at least it seems to do it for me.

[Frederf]

This is such a confusing issue. First, raise your hand if you know what the letters S.A.S. stand for. It's stability augmentation system. What does it mean to augment stability? Usually such systems are effectively dampers designed to suppress unwanted oscillations and not autopilots. Historically KSP SAS & ASAS were exactly autopilots which instantly starts everyone off on the wrong foot in terms of expectations. We've departed traditional vernacular at right angles already months ago.

Previous versions were a motley collection of controls, control channels, and controllers. "SAS" parts were controllers only available to the control channel used by the "T" SAS controller. The manual controller actuated all available controls except the SAS torque which was reserved for hands off input. The ASAS simply expanded the set of controls that the SAS controller had available to it beyond pod torque and the exclusive SAS wheel parts. To top off the confusion you could have multiple controllers in operation simultaneously with manual control channels operating one set of controls and the automatic SAS channel(s) operating a partially overlapping set. What a mess!

In any case all automatic inputs were effectively PID controlled. Things wobbled with any sort of automatic input because 1. the PID coefficients were terrible and 2. they were static.

---

What does the new system do? Surely it is a PID contoller as well. My fear is that despite making "better coefficients" the PID factors are still static and thus unsuitable beyond some range of use cases. My hope would be that the PID coefficients would be autotuned if not outright fully configurable manually. Are the PID numbers autotuned or not?

---

Ultimately the typical user will want both stability enhancement for maneuvers and autopilot functionality for non-maneuvers. Before we had A, it seems now we have B. Living without both available is going to be unsatisfactory.

Edited July 25, 2013 by Frederf

[Nodrak]

https://www.dropbox.com/s/frj8dyqbvb2i0oh/_21%20SAS%20Test.craft

Here is an example rocket.

The first stage it looks like the 4 fins are just not cutting it. Probably working as intended.

The second stage on the other hand, with 4 gimbaling engines has plenty of controlability when flown manually. If you fling the rocket a certain direction you can easily slow down the movement and correct it by imputing controls in the opposite direction.

But the SAS can't. It sluggishly does do it eventually when movement has slowed to a certain point (while rolling slightly) but clearly not as easily as manual correction. Maybe the SAS is not using gimbaling enough and is relying on capsule torque.

Because the last stage works wonderfully and if it was always like that then I think everyone in this thread would be happy.

Downloaded, tested. Flew fine. First 2 stages have slight instability from their design (ie, old SAS would have destroyed that ship if it was heavier). I didn't turn with the SRB only stage, because well that's crazy, you would rely on your Pod Torque to counter drag. Fins on the end of a ship, (like an arrow) are for flying straight, not turning and maneuvering. Once you drop the Fins, all roll control is from your Torque, you added an inline which shows when you punch it in Stage 2, which would roll out of control in .20 ASAS no doubt.

For a craft that is basically 4 stilts with rockets on the end of them, the SAS did wonderful, far better than than it would have in .20

[HarvesteR]

Well this is weird, how do some people have working SASs and others not? I know, how about we all post pictures of the ships we are using to find out if there is a problem or not. Because I'm convinced that there really is something wrong.

Now this is much more like it!

I'm also suspecting from all the reports that something, could be a mod, could be a bad settings file, could be old parts lingering about the KSP/Parts folder, might be interfering with normal SAS operation.

The new SAS might be a lot of things, but it most definitely isn't a thing that requires constant input to maintain attitude. It's been pretty consistently doing that on all the test cases we did here, and there were a lot of them.

If the SAS isn't holding attitude properly for you, then it very well may be in a buggy state, because that is certainly not the intended behaviour.

First things first though, we need a proper baseline to compare different cases here. So how about we use the Aeris 3A spaceplane as a benchmark? It's easy to fly and it's got a few particular flight characteristics that are easy to spot.

So, the test is as follows.

* Launch a new Aeris 3A from the runway, set it at full throttle and try to maintain level flight.

* Do not adjust trim. On neutral trim, the 3A has a slight tendency to pitch down when the tanks are full.

* Climb out to 1000m altitude and verify that it's doing that. Let go of the controls and make sure this tendency to drop the nose is there.

* Ok, that's our control. If you've observed anything else at this point, something is already wrong with your game and you should be doing a complete reinstall.

* Now, let's test the SAS. From level flight, engage it with T and don't apply any input. Verify that it's maintaining attitude.

* If that didn't happen, then something is wrong with your game. Reinstall.

* Now, a few more tests. Apply some slight pitch up input without disengaging SAS. Pitch up to about 30° and watch the aircraft maintain the new attitude.

* If that didn't happen, then something is wrong with your game. Reinstall.

If you spotted something wrong by doing this test, please list all your mods if any, and check the KSP/Parts folder to see if there's anything in there (on a stock install it should be empty). Also, try deleting your settings.cfg file and running the game again. It's unlikely, but not impossible that a bad config would mess things up.

That should be enough to spot most cases where things are very obviously wrong. We can start with that for now.

Cheers

[carazvan]

Added a new set of pictures at a lower altitude ... some problems might be amplified at higher altitudes (here it actually looks like the opposite with the 45* climb being worse at lower altitude) ... but I will grant you the level to 20* seems to be a bit better .. maybe only 15* or so instead of 10*

If you are trying to compare .21 and .20 with those pictures, you should probably make sure they happen within the same magnitude of atmosphere... All this is showing is the obvious? Space plane in low atmo without RCS having slightly harder time controlling pitch in plane that wants to pitch down. AKA, you have less magic torque.

I just flew to the moon with Kerbal X to test, works like a charm even though Kerbal X is tiny off balance. Only issue is SAS is slow to respond to manual input, as you noticed. Nothing crippling, and actually works exactly how the OLD ASAS did, you just expected it because its this T button you hit...

[Feldynn]

I think what I'm seeing with the new SAS system is a general inconsistency across different craft, it seems like there are differing limits to how effective it is based on the size of the craft, amount of torque and thrust. For the first time ever I tried a few of the stock space planes, the bigger Ravenspear ones seemed generally to be the least likely to throw a fit I think because they had a decent thrust / weight / torque ratio.

I tried the Aries 3A too and while it was a little twitchy to begin with but wasn't too bad once I got used to it (again, first time flyer here), it's still got a reasonably thrust / weight / torque ratio but is obviously a lot lighter so it can be twitchy and was sticking to it's course like glue once I stopped messing with it. I did manage to send it into a SASsy spin cycle / face plant a couple of times but that was mostly me turning too hard.

I think the worst plane I tried was the Raven Mk1, not a huge plane but it has 3 engines (two turbo and one basic I think) which just seemed like too much and would cause the SASsy flippage a lot while trying to turn. Shutting down the central engine seemed to help somewhat, it would still wig out on occasion but it was much easier to control with just the 2 engines running.

Rockets are a bit different again but still show a similar kind of inconsistency, at certain stages they seem to hold course with SAS fairly well (I found landing on Mun MUCH easier with new SAS though it could have been a little more snappy about holding the course once I ceased control input) but at other times not so much and I'd find I was making far too many manual corrections.

With the old ASAS it was a very singular system, if ON then you're on "this" heading and that's the end of it for the most part and if OFF then you drift / spin around freely until you do something, I actually like the way the new system works when you find something it works well in but it just doesn't seem to have the consistency across a multitude of craft that the old system had and I think that's what's causing a lot of the distress right now.

[Naelo]

Now this is much more like it!

I'm also suspecting from all the reports that something, could be a mod, could be a bad settings file, could be old parts lingering about the KSP/Parts folder, might be interfering with normal SAS operation.

[snip]

The Aeris 3A has a slight tendency to pitch up in my install. The ASAS takes ~5 seconds to lock into a new heading and doesn't really compensate for the upward pitch. I'll redownload full from the store and report afterwards.

[Pingonaut]

I take back what I said about liking it. I'm having to micromanage my ships to keep them stable, and I haven't tried, but docking must be a huge pain in the butt, now. I'll reinstall my game and see how it goes.

Edited July 25, 2013 by Pingonaut

[RocketPilot573]

Err, why does the Aeris 3A have a functioning SAS when nothing else does? I'll try re-downloading tomorrow.

[Naelo]

Quick side-question: Is the Avionics package now the same SAS as the ASAS? (Excluding reaction wheels of course)

[Nodrak]

3A in a fresh .21 off the store is slightly pitching up after taking off and doing 180 back over KSC. By slightly I mean, 30s for 1 pixel on navbal. If you take off from the runway asap and fly straight, it noses down slightly, but is nose up slightly by the end of the runway. SAS off, landing gear up/down same, torque on/off same.

[naloxone]

Be sure to read the pod descriptions. Only Manned pods are "SAS equipped" by default. So if you have a probe controlled craft, the SAS functionality will do nothing, just like 0.20 ships without ASAS installed. If you want SAS to do anything with a probe body, then you have to add one of the other SAS equipped parts, which include the Old ASAS 1m & 2.5m parts, or the Avionics nosecone thingy.

Say again? If we want "SAS" we are supposed to add parts that used to be "ASAS"?

[EdisonTrent]

Say again? If we want "SAS" we are supposed to add parts that used to be "ASAS"?

Parts that have SAS-Equipped are the parts that give you new ASAS and Contains Reaction Wheels are the new SAS. (New as in .21)

[Tiron]

Now this is much more like it!

I'm also suspecting from all the reports that something, could be a mod, could be a bad settings file, could be old parts lingering about the KSP/Parts folder, might be interfering with normal SAS operation.

The new SAS might be a lot of things, but it most definitely isn't a thing that requires constant input to maintain attitude. It's been pretty consistently doing that on all the test cases we did here, and there were a lot of them.

If the SAS isn't holding attitude properly for you, then it very well may be in a buggy state, because that is certainly not the intended behaviour.

First things first though, we need a proper baseline to compare different cases here. So how about we use the Aeris 3A spaceplane as a benchmark? It's easy to fly and it's got a few particular flight characteristics that are easy to spot.

So, the test is as follows.

* Launch a new Aeris 3A from the runway, set it at full throttle and try to maintain level flight.

* Do not adjust trim. On neutral trim, the 3A has a slight tendency to pitch down when the tanks are full.

* Climb out to 1000m altitude and verify that it's doing that. Let go of the controls and make sure this tendency to drop the nose is there.

* Ok, that's our control. If you've observed anything else at this point, something is already wrong with your game and you should be doing a complete reinstall.

* Now, let's test the SAS. From level flight, engage it with T and don't apply any input. Verify that it's maintaining attitude.

* If that didn't happen, then something is wrong with your game. Reinstall.

* Now, a few more tests. Apply some slight pitch up input without disengaging SAS. Pitch up to about 30° and watch the aircraft maintain the new attitude.

* If that didn't happen, then something is wrong with your game. Reinstall.

If you spotted something wrong by doing this test, please list all your mods if any, and check the KSP/Parts folder to see if there's anything in there (on a stock install it should be empty). Also, try deleting your settings.cfg file and running the game again. It's unlikely, but not impossible that a bad config would mess things up.

That should be enough to spot most cases where things are very obviously wrong. We can start with that for now.

Cheers

What throttle setting is this 'nose-down tendency' at, since you don't mention throttle at all. On mine, it holds almost perfectly stable at full throtle, and gains a nose-down tendency when the throttle is cut. The SAS *is* able to hold...eventually.

I've admittedly only tried to use the stock SAS once since the update, and I experienced pretty much what they were complaining about: Constantly having to make adjustments to keep doing what I was trying to do (point prograde). Admittedly my target was probably moving slightly(Prograde in the middle of a maneuver), but what it seemed like was happening was that when I released the controls while still moving, it would take several seconds to start trying to stabilize that movement. And it would frequently seem to stabilize at a point a short distance AFTER I let go. It DID seem to be stabilizing, but not where I wanted it to, and there did seem to possibly be some drift, necessitating a lot of manual corrections. It wasn't long before I got frustrated and hit 'prograde' on the SMART A.S.S. again.

The craft in question had very low torque and no other control surfaces, but there's a slight element of this visible in the Aries3A behavior too.

Edited July 25, 2013 by Tiron