Come back old ASAS - all is forgiven!

[Fernbhoy]

Another video test to see if it helps devs/users out

this is my satellite launcher from 0.20 rebuilt for 0.21 with the Adv SAS included. this rocket had no trouble getting into orbit last patch

As you can see it has no trouble for the first 15 secs of launch then it falls apart (literally)

notes: yes i forgot to strut 2 rockets to the centre rocket

yes i forgot to add my rcs boosters

[Whackjob]

Fernbhoy, the problem with that rocket has nothing to do with ASAS. That rocket is held together with tissue paper and spit. You've got a counterweight at the top that's free to wiggle around, and it's putting a torque on the rest of the rocket. Then you get a resonance going that tears the bottom apart.

Add more struts. Secure the top as well. I suggest putting a large truss on top the upper orange tanks, and send struts from there to the top component. Brace that thing!

[Comrade Jenkens]

It took a bit of getting used to but i'm loving the new SAS. However it needed reaction wheels, RCS or control surfaces to work effectively now.

[purpletarget]

OK, here's an updated video of my problem for those apparently wanting to attribute a sudden 10º shift in heading to orbital precession. To eliminate that possibility, I rolled the ship 90º. I also selected a much higher orbit.

Interestingly the effect seems less pronounced in a higher orbit (no, it's really not orbital precession I'm seeing). This video is much shorter. I thrust from stationary, the ship pitches up about 5º and hods it there (briefly - it runs out of fuel at this point), and then when the engines cut out it returns to the set heading.

Your new vid says the same thing...your SAS is working fine.

The orbital procession in your original video explained why SAS returned your final heading (after engines were shut down) to a position that was 10-12 degrees above the horizon, which is where the test was started a minute beforehand.

As for the 5-10 degrees when thrusting, that's because, as you originally noted, your craft is unbalanced.

However, the new behavior of the SAS doesn't try to fight all the way back to your locked heading in that case. All it does is trim out the controls until it can stop the yaw/pitch and holds it there. (Assuming the craft has the control authority to do so) Unlike the Old ASAS, it doesn't (as I also already mentioned) go and fight it's way back there like a dog with a bone. If it tried, it would over-correct, and go past the set heading (as we see it gently do once the thrust is released). That over-correction, if allowed under thrust, is the kind of thing that resulted in oscillations and the shakiness that was the bane of the old ASAS.

A key difference in flying the new SAS is that it cuts out when manipulating the controls to allow the freedom of maneuver, but if you want a particular attitude/heading, you need to kill most of the swing yourself, or let off the controls early enough that the SAS can kill the swing once it steadies up on your intended course. So unlike before, it doesn't lock on the heading that you let go...but rather it locks up to hold the attitude once it steadies up.

[rytheguy]

Your new vid says the same thing...your SAS is working fine.

The orbital procession in your original video explained why SAS returned your final heading (after engines were shut down) to a position that was 10-12 degrees above the horizon, which is where the test was started a minute beforehand.

As for the 5-10 degrees when thrusting, that's because, as you originally noted, your craft is unbalanced.

However, the new behavior of the SAS doesn't try to fight all the way back to your locked heading in that case. All it does is trim out the controls until it can stop the yaw/pitch and holds it there. (Assuming the craft has the control authority to do so) Unlike the Old ASAS, it doesn't (as I also already mentioned) go and fight it's way back there like a dog with a bone. If it tried, it would over-correct, and go past the set heading (as we see it gently do once the thrust is released). That over-correction, if allowed under thrust, is the kind of thing that resulted in oscillations and the shakiness that was the bane of the old ASAS.

A key difference in flying the new SAS is that it cuts out when manipulating the controls to allow the freedom of maneuver, but if you want a particular attitude/heading, you need to kill most of the swing yourself, or let off the controls early enough that the SAS can kill the swing once it steadies up on your intended course. So unlike before, it doesn't lock on the heading that you let go...but rather it locks up to hold the attitude once it steadies up.

So if it requires you to kill most of the swing yourself, what's the point of not flying manually? Also when is "letting off early enough"? That doesn't seem intuitive at all, and why make it weigh more when it's a bit less useful than it was before?

[DMagic]

I think Whackjob is right Fernbhoy. That rocket may have been stable before, but it probably shouldn't have been. That cupola is just bouncing around up there, and those two radial engines don't look stable at all, the force of their engines is pushing them off center long before the rocket explodes.

The rocket also has very little in the way of torque; I think that capsules only provide torque when manned, so only your command pod is doing anything. The 2.5m grey ASAS module is completely redundant in your situation, it provides no torque and the command pod provides SAS functionality. The only use for this part is on large probe controlled crafts (and even this won't be true if they add SAS to the probe cores).

[FleetAdmiralJ]

The main problem I'm still having with the SAS is on the gravity turn. I'll turn it, it will sling back to where it was, then slowly move back to where I let go. I would think it should just say where I let it go, especially if there is little momentum that could cause it to fling around. I've had to hold the S button down to make sure it stays in the turn until the SAS can "catch up" to it a couple of times.

However, other than that, it's been working like a charm on rockets. No problem launching straight up, no problem keeping a heading when I burn.

[purpletarget]

So if it requires you to kill most of the swing yourself, what's the point of not flying manually? Also when is "letting off early enough"? That doesn't seem intuitive at all, and why make it weigh more when it's a bit less useful than it was before?

SAS is not a mind reader....it'll hold the attitude you set, or trim out the craft to hold as best it can if unbalanced. But you need to tell it what heading you want... so it's either the heading you're on when you turn it on, or if you're fiddling with the controls, it'll be where'ever the ship steadies up. So if you want a specific heading, you probably want to make sure there's not much swing when you get the nose onto it.

If you drive anything that tends work in fluids, this manner of control does make intuitive sense...boats at low speed require similar attention for example.

And the SAS doesn't weigh anything...it's a computer. Most of the mass of control parts is mostly to do with the reaction wheels. (which are a electricity to torque converter, not a stabilizer....separate issue)

[regex]

However, the new behavior of the SAS doesn't try to fight all the way back to your locked heading in that case. All it does is trim out the controls until it can stop the yaw/pitch and holds it there.

Okay, but it doesn't respond to course corrections after that. For instance, my unbalanced craft slides off the maneuver node and gets trimmed out, then I pull it back to the maneuver node, then the SAS lets it slide right back to the trim point. Over and over. It's maddening.

So unlike before, it doesn't lock on the heading that you let go...but rather it locks up to hold the attitude once it steadies up.

That, quite frankly, makes it completely useless. I might as well not even bother installing it on a craft, if I can have to fly it manually better anyway. Looks like I'll be joining the proud ranks of MechJeb users for its KILLROT function.

E: What makes it even more maddening is that the SAS remembers the original heading and points the craft there once the burn is done. Why even bother correcting back to the origin if the intention is to ignore it when it matters?

Edited July 25, 2013 by regex

[karolus10]

I can agree with people above, sometimes rockets are drifting slightly off course when ships aren't balanced ,but it seems like SAS wont fully use engine gimbals and RCS to compensate rotation if necessary.

Also I found quite annoying for SAS turning off every time when switching between rotation and translation modes - new SAS allow for smooth rotation making this feature obsolete and troublesome, when switching between rotation and translation in quick succession .

Edited July 25, 2013 by karolus10

[Whackjob]

For what it's worth, I do notice that the ASAS does tend to wander, but I think it's supposed to be like that. That is my impression. Before, you'd be locked into a fine direction, and your gimbals and control surfaces would go nuts just keeping that rocket dead on that point. Your RCS fuel got eaten alive as a result. Now, it'll keep your rocket headed in a general direction, but wander a bit within this circle. Even with my giant constructions, I see that wandering, but when it strays slightly to the side, it does make a correction.

So I think we sacrificed that total precision for a more realistic approach towards attitude control. We conserve RCS and we're still heading in the right direction within a degree or two.

That being said, the aerospike engines are totally useless. I've experimented with those, and for whatever reason, using those at anything past one third thrust causes you to completely lose control, ASAS activated or not.

[Giggleplex777]

Sean's right. Quickly pitching up in the Aeris 3A will cause it to point at retrograde, and a tiny amount of input will return the plane to prograde. The longer you fly the plane, the more this happens.

Can someone make a video of this happening?

[Giggleplex777]

For what it's worth, I do notice that the ASAS does tend to wander, but I think it's supposed to be like that. That is my impression. Before, you'd be locked into a fine direction, and your gimbals and control surfaces would go nuts just keeping that rocket dead on that point. Your RCS fuel got eaten alive as a result. Now, it'll keep your rocket headed in a general direction, but wander a bit within this circle. Even with my giant constructions, I see that wandering, but when it strays slightly to the side, it does make a correction.

So I think we sacrificed that total precision for a more realistic approach towards attitude control. We conserve RCS and we're still heading in the right direction within a degree or two.

That being said, the aerospike engines are totally useless. I've experimented with those, and for whatever reason, using those at anything past one third thrust causes you to completely lose control, ASAS activated or not.

That's caused by its lower-than-usual drag value (0.1 instead of 0.2).

[rytheguy]

SAS is not a mind reader....it'll hold the attitude you set, or trim out the craft to hold as best it can if unbalanced. But you need to tell it what heading you want... so it's either the heading you're on when you turn it on, or if you're fiddling with the controls, it'll be where'ever the ship steadies up. So if you want a specific heading, you probably want to make sure there's not much swing when you get the nose onto it.

If you drive anything that tends work in fluids, this manner of control does make intuitive sense...boats at low speed require similar attention for example.

And the SAS doesn't weigh anything...it's a computer. Most of the mass of control parts is mostly to do with the reaction wheels. (which are a electricity to torque converter, not a stabilizer....separate issue)

I don't understand how it holds the attitude I set, when the idea behind setting a certain attitude is "let go a little early so you can go to the right place." What is a little early? How long to I hold D to set an attitude of 70 degrees at an 090 heading without my SAS overshooting? Where's the predictability or intuitiveness? Why take a system that was predictable and usable and replace it with guess work? What's the point on having it if I cant use it in a predictable manner?

And I was referring to the mass of the reaction wheel as part of SAS because the pods, while they do seem to have some torque according their descriptions, probably don't have enough to maintain heading with a larger rocket, so I was kind of thinking the reaction wheel was a given for full functionality on the new SAS system. Sorry for being confusing.

[Fernbhoy]

I think Whackjob is right Fernbhoy. That rocket may have been stable before, but it probably shouldn't have been. That cupola is just bouncing around up there, and those two radial engines don't look stable at all, the force of their engines is pushing them off center long before the rocket explodes.

The rocket also has very little in the way of torque; I think that capsules only provide torque when manned, so only your command pod is doing anything. The 2.5m grey ASAS module is completely redundant in your situation, it provides no torque and the command pod provides SAS functionality. The only use for this part is on large probe controlled crafts (and even this won't be true if they add SAS to the probe cores).

so remove the ASAS all together? so for a rocket like that should i have any of the controls built in bar RCS?

also what are Truss? the scaffolding?

[Whackjob]

Really? So if I protrude something below with high drag.... hrm. Time to play around a bit...

[MonkeyLunch]

I vote for old system. At least it did something before.

[BubbaWilkins]

Okay, but it doesn't respond to course corrections after that. For instance, my unbalanced craft slides off the maneuver node and gets trimmed out, then I pull it back to the maneuver node, then the SAS lets it slide right back to the trim point. Over and over. It's maddening.

That, quite frankly, makes it completely useless. I might as well not even bother installing it on a craft, I can fly it manually better. Looks like I'll be joining the proud ranks of MechJeb users for its KILLROT function.

E: What makes it even more maddening is that the SAS remembers the original heading and points the craft there once the burn is done. Why even bother correcting back to the origin if the intention is to ignore it when it matters?

Fix your balance. The old system used brute force through pod torque to make crafts behave. On smaller craft, you could be grossly out of balance and still not really have any issues. The new system is much more subtle and requires your craft to be near neutral balance wise. Trim should only adjust mildly to compensate for changes in speed, CoM due to fuel consumption, or lift due to altitude.

[Brotoro]

Sean's right. Quickly pitching up in the Aeris 3A will cause it to point at retrograde, and a tiny amount of input will return the plane to prograde. The longer you fly the plane, the more this happens.

Can someone make a video of this happening?

Deactivate fuel flow from the front tank. Then the plane will be LESS likely to pitch up violently as time goes on. Just remember to reactivate the front tank just before the rear tank runs dry.

[Sean Mirrsen]

Sean's right. Quickly pitching up in the Aeris 3A will cause it to point at retrograde, and a tiny amount of input will return the plane to prograde. The longer you fly the plane, the more this happens.

Can someone make a video of this happening?

Already did.

Okay, so I reinstalled and made another test with the Aeris 3A. I've no idea how to Youtube, so here's the video on Dropbox.

https://www.dropbox.com/s/crr6am98zh5a91p/KSP%202013-07-25%2021-04-58-647.avi

It appears I've been wrong, it is possible to make the Aeris flip out with some intensive abuse, but it's a lot easier with SAS on. I should probably remind Squad that ASAS doesn't stand for Aerial Stunt Augmentation System.

If you can toss it up on Youtube so people can see it easier, it'd be quite helpful.

The flipping thing with the Aeris is really just in the aerodynamics of it. When you pitch up like that and release, SAS immediately throws the pitch into the other direction - but at that attack angle, all it does is flip the thing over. You can replicate the effect manually, but it takes some doing - and SAS treats it like a perfect "reverse course" maneuver, sticking to the retrograde direction. It also loses track of where you were trying to point it if it flips back on its own.

Edited July 25, 2013 by Sean Mirrsen

[Inglonias]

After using the new SAS for a bit, I'm finding that it's not holding course much either. The videos said that it only releases control of axes which you are controlling, but I find that my rocket tends to roll when doing a gravity turn with SAS on. Not sure if it's supposed to do that or not.

[regex]

Fix your balance. The old system used brute force through pod torque to make crafts behave. On smaller craft, you could be grossly out of balance and still not really have any issues. The new system is much more subtle and requires your craft to be near neutral balance wise. Trim should only adjust mildly to compensate for changes in speed, CoM due to fuel consumption, or lift due to altitude.

What is the new SAS for, then? If it can't correct for minor imbalances in the craft during a burn then I'd hesitate to call it a stability system. If it remembers the correct attitude but doesn't attempt to maintain it until all forces are neutral, then it's pretty useless for maintaining attitude.

If I built a perfectly balanced craft I wouldn't need the SAS in the first place. So, again, what's it for? HarvestR seems to think it's for attitude maintenance and I agree. The problem is that it's not doing that right now.

If the current functionality is the intended behavior then I'll just shut up and use MechJeb, but seeing as how two devs have already acknowledged that this is an issue I have problems believing this is the intended behavior.

[TMS]

I can't say that I've had the same extent of problems that others have, but a couple of things are obvious after some basic testing:

The new SAS is much more intuitive and helps vertical landings immensely. I just performed the most casual and controlled moon landing I've ever done. Most landings are an exercise in switching ASAS on and off and compensating for my own overreaction. Not any more. The techniques of following the retrograde marker onto the radial point work easily with the new SAS system and allow one to concentrate on thrust and pitch without having to look at the keyboard for the T-key. I practically did this one-handed. Brilliant feature and must not be lost.

However, it also clear that a balanced rocket that can perform a gravity turn perfectly (with SAS holding two axis still, while 90 degree heading is achieved) will become completely uncontrollable if even the smallest of fuel tanks is added to make it asymmetric. Movements on one axis will result in all manner of movements on the others. This will account for some of the complaints - especially those who specialise in building monstrosities. The weird thing is... that it can hold attitude once you stabilise it, but it's an unpleasant gameplay experience.

The gimbals aren't being fully used to compensate for imbalance, despite there being enough authority to do so. People need to stop ignoring that fact as a means of bashing others about their designs.

On a side note, it does make me wonder how this got through the QA group. Are changes and features listed to the QA group and they work through them systematically, judging them against a clear description of how something is supposed to work? I wonder whether nobody questions the behaviour of features as they don't want to appear too critical and risk their spot on the testing group? Meh... one for another day.

I'm sure it'll get fixed by Chads brain... and the huge manatees will swim into the distance.

[Benie]

I'll just shut up and use MechJeb

I'm sorry, dude. But someone posted that MJ's Ascent Guidance is bugged. Using an unmanned probe, the nav ball was pointing to the North. Without it, it points fine.

He never messed with the probe. He even has a screenshot.

[0xFADE]

I do hope something gets resolved. I thought the old asas was a pain to use till it was gone.