New SAS functionality and You! [0.23 Update]

[DMagic]

0.23 Update

Not much changed in 0.23. Subjectively the SAS system feels about the same, there may be some minor tweaks from 0.22, but I haven't noticed it.

No changes to the parts.

There are two issues that do deserve mention.

The "docking mode" control bug has been fixed. Using the WASD keys in docking mode for translation control will no longer activate damping mode. The roll controls (QE) do activate damping mode, but this is to be expected.

While not strictly related to the SAS system, the new tweakables system can affect stability. Maximum engine thrust can now be controlled separately for each engine. In the VAB you can monitor the center of thrust during such adjustments to account for mass imbalances on a non-symmetrical craft. Thrust can also be altered in-flight, but obviously you won't be able to directly observe the new center of thrust.

Everything else below should still be valid for the current SAS system.

---------

0.22 Update

There have been some changes in the 0.22 update to the SAS system.

Parts:

Inline Reaction Wheel - No change

Inline Advanced Stabilizer - No change, provides the same torque as the inline reaction wheel but has a higher mass; don’t use this part

ASAS Module (2.5m grey wheel) - No change, still has the same aggravating name, still provides the same torque as the smaller reaction wheels at a lower weight

Avionics Package ---> Sensor Array Computing Nose Cone - No longer has any SAS function; this is now a science system part

Command pods and probe cores - No change

Hex Probe Core - Still has bugs in .cfg file; includes old, pre-0.21 SAS entries and bounded by ')' instead of '}'.

MkI Command Pod - Has the same amount of torque, but doesn't seem to shake as much with SAS on.

Functionality Changes:

Summary:

- SAS indicator on Navball – Now shows when you are in damping mode by replacing “SAS†with two spinning arrows icon

- Higher usage of available control authority – Can maintain heading better, but may cause more oscillations

- Unbalanced crafts maintain heading better under thrust – Should see less of that “stable drift†that happened in 0.21 – See below #

- Damping mode – Switching to damping mode briefly causes controls on all axes to drop to zero – See below &

- RCS translation controls - Damping mode activated by translation when using 'docking mode', but not when using 'staging mode' - See below *

System Changes:

There are several aspects of the SAS system that have been tweaked in 0.22. The first, and most visually obvious change is the SAS indicator above the Nav Ball. When SAS is activated with “F†or “T†the light turns on as normal, but when the system is in damping mode the indicator changes to two spinning arrows. Any manual input will switch the system to damping mode; it will stay in this mode for a brief moment after releasing manual input (how long it stays in damping mode depends on how much the craft is changing direction). Once the system is in locking mode the light will change back to the “SAS†indicator.

The system still works off of your true orientation in the Kerbal universe. This means that while SAS is activated you will stay oriented in the same direction with respect to the universe, regardless of your position with respect to the planet/moon you are orbiting. This is effectively the same thing that happens when you activate timewarp. If you point at the prograde vector, wait half an orbit and you will be pointed retrograde; if you point at the surface, wait half an orbit and you will be pointed directly up from the planet. If you create a maneuver node and point at the blue vector you will follow it throughout the orbit (though you will slowly drift away from the vector, I assume this is due to the orbit of an object around its parent body; this doesn’t happen in Kerbol orbit).

On the whole, the SAS system seems more aggressive. It uses more available control authority to try to maintain heading. This can cause more oscillations if you aren’t careful. By using manual corrections and toggling off/on SAS with “F†it is usually possible to avoid these large oscillations. This isn’t always easy though, and very long, or otherwise unstable crafts will probably have this problem regardless of what you do; that is more of a problem of bad craft design than a faulty SAS system (that some of these designs may have worked in previous versions doesn’t mean they aren’t problematic).

Roll control has been greatly improved. In many cases SAS will stop roll almost immediately after control is released; with relatively small crafts it works almost as well as using RCS to stop rolls. You can still overload its ability to prevent roll (SRBs are great for doing this, which might cause some issues early in career mode), but it is better overall.

# - Unbalanced crafts should work better in 0.22; this is best demonstrated with a test case. These crafts are identical, but were built from scratch on new saves in both 0.22 and 0.21. Click on the pictures for bigger versions. The craft has one command pod, one large reaction wheel and one gimbaling engine. During launch it is easily able to handle the imbalance in both versions, though 0.22 does feel more stable overall.

Once the upper stage is separated the behavior of SAS differs between the two versions. In 0.21 SAS can hold the craft steady at up to about 50-60% throttle, after that the craft will begin spinning out of control; in 0.22 it is stable at a slightly higher throttle. In both cases SAS is using full control authority on the pitch and yaw axes to maintain heading.

There are two important points here, and one big difference. In 0.21, even when the craft is held steady it won’t maintain its original heading. It drifts a few degrees in the direction of the imbalance, then holds its heading there, even when there is excess control authority available. In 0.22 this drift is greatly reduced; SAS mostly holds the original heading until it loses the ability to maintain heading at all (this is greatly helped by slowing increasing throttle, instead of going straight to the max).

One important point that can be observed here is that it is possible to get stuck in damping mode. If you start rotating or go beyond the system’s ability to maintain heading the indicator will switch to the damping icon and will stay there. This doesn’t seem to affect the amount of control authority used, it simply means that the system has no heading to lock onto.

& - The other point is that damping mode doesn’t work the way you might think. At around 50% throttle the controls on the pitch and yaw axes are almost maxed out and the system is in locking mode. If you then rotate, the system switches into damping mode. Right away, you will notice that all control on the pitch and yaw axes drop to zero. This will cause the craft to start spinning out of control despite the fact that there is enough control authority to hold it steady. The system will resume controlling the pitch and yaw axes (even if you continue to hold the rotate controls), but that brief interval where the controls drop to zero is enough throw off the system.

RCS:

* - Using the translation controls for RCS in 'staging mode' (the H J K L I N keys) will not cause damping mode to be activated. Using the translation controls in 'docking mode' (the WASD keys) will cause damping mode to activate. This remains the case with 'fine controls' (activated with the CAPS lock key) on or off, and this also occurs even with RCS turned off. This behavior is not entirely predictable, activation of damping mode seems to be at least somewhat dependent on whether or not the RCS thrust is imbalanced. And excessive input, or very imbalanced thrust can cause damping mode to activate in 'staging mode', but I believe this is because the craft is momentarily unable to hold its heading (similar to the situation described above), and is not the same as what happens in 'docking mode'.

The 'staging mode' case seems to be more useful, docking usually requires careful control over your heading, and damping mode is not as effective as locking mode for providing this. I would recommend staying in 'staging mode' and using the translation controls instead of using 'docking mode'.

Will update as more info becomes available.

0.21 Information

Most of the information below should still apply.

See C7's latest blog post on the new SAS system

The 0.21 update brought with it many changes to the SAS functionality and a whole host of complaints to go along with them. So I feel like I should attempt to explain how some of these new systems work.

Keep in mind that this is from my non-developer point of view, so if you find anything blatantly wrong here just tell me and I'll update this post. That said, most of what's changed can easily be understood by taking a minute to look at the part .cfg files.

Changes from the old system:

ASAS functionality is gone, replaced by SAS

Command pod torque is replaced by the reaction wheel system

Part names and functions:

SAS Module (1.25m grey wheel) ---> Inline Reaction Wheel - Provides torque and SAS function

ASAS Module (1.25m yellow wheel) ---> Inline Advanced Stabilizer - Provides torque and SAS function

ASAS Module (2.5m grey wheel) ---> Same name - Provides torque and SAS function

Avionics Package ---> Same name - Provides SAS function only

All command pods, cockpits, and probe cores ---> Same names - All provide varying amounts of torque, and SAS function

New System:

Reaction Wheels:

These are what provide rotational torque. They allow your crafts to move without using gimbaling engines, RCS, or control surfaces in an atmosphere. This is an improvement on the old system and is more realistic than the magical command pod torque.

Pushing "T" to turn on SAS has no effect on the reaction wheel system. You can produce torque regardless of whether or not the SAS light is on.

All parts with reaction wheel functionality use electrical power when you turn, they all have an option to toggle torque if you right-click on them, and they all have the line "name = ModuleReactionWheel" in the .cfg file, which is followed by the specific torque values in each axis and the electrical power consumption rate.

All command pods, cockpits, probe cores, and the three reaction wheel parts mentioned in the section above provide this function. The chair, however, does not.

SAS Functionality:

The new implementation of SAS is probably the more confusing change. This is what tries to maintain your heading (true orientation actually, which will change as you orbit a planet), controls reaction wheels, activates control surfaces, gimbaling engines, and RCS thrusters. Any and every control system is controlled by any part with the SAS function. You do not need to add an Inline Advanced Stabilizer or Inline Reaction Wheel to control RCS, gimbals, or control surfaces.

SAS is not nearly as aggressive as the old ASAS functionality. This reduces wobbling and over corrections a lot, but in many cases your rocket might drift away from its current heading. The 0.21.1 update addresses this and seems to reduce or eliminate drift in most cases. Adding more Reaction Wheel parts can help eliminate drift or accommodate an unbalanced rocket.

As detailed in C7’s blog post linked above, there are two modes for the SAS system. In damping mode, SAS will not try to maintain a heading, it will only provide stability and should allow for smoother control over your craft. In locking mode SAS will try to maintain a heading. There is, however, a delay in switching between damping and locking modes. Constant manual input (or joystick signal noise) will prevent locking mode from activating.

This is not an autopilot, so don't expect it to be one. It also does not lock out the controls, you can activate SAS and still control your craft.

As mentioned above, it is present in all of the manned command pods, cockpits, all of the probe cores, the Inline Reaction Wheel, the Avionics Package, the 2.5m grey ASAS module, and the Inline Advanced Stabilizer. The chair, again, does not have SAS functionality or provide torque.

Any part with SAS functionality will have the line "name = ModuleSAS" in the .cfg file. This function does not use any electrical power, it has no right-click toggle value, and it does not provide any rotational torque by itself.

When you push "T" (or hold "F") to turn on SAS, you are now activating the new SAS functionality. There is only one effect of activating SAS now, not like before where you had the somewhat conflicting SAS/ASAS functionality activated with "T".

RCS Changes

While not strictly related to the new SAS system, there has been an important change to the way RCS functions. When under the fine control mode (activated with CAPS Lock, and indicated by the icons in the lower left corner changing from orange to blue) the RCS thruster compensate for their distance from the center of mass. Thrusters further away from the COM will provide less thrust. This should help to reduce unwanted rotation while docking. Don’t expect it to work perfectly; proper RCS placement is still important.

.cfg File:

Here is an example of the relevant portions of the Mark1-2 Command Pod .cfg file:

MODULE
{
	name = ModuleSAS
}

MODULE
{
	name = ModuleReactionWheel

	PitchTorque = 15
	YawTorque = 15
	RollTorque = 15

	RESOURCE
	{
		name = ElectricCharge
		rate = 1.2
	}
}

“name = ModuleSAS†provides the SAS function.

“name = ModuleReactionWheel†and the lines following it provide torque.

It should also be noted, that while the old ASAS functionality has been removed from all parts in KSP, it is still present and can be reactivated. Adding the line “module = AdvSASModule†to a part’s .cfg file should provide the ASAS function. See UmbralRaptor’s post down the page a little for more on this.

Usage in 0.21:

So there is now a change in the proper usage of these parts. The only purpose of the reaction wheel parts is to provide extra torque; the 1.25m yellow Inline Advanced Stabilizer, the Inline Reaction Wheel, and the 2.5m ASAS parts. The command pod or a probe core should provide any SAS functionality that you need. Again, chairs present a special use case and will require either a probe core or one of the parts listed above, both to provide SAS functionality and to provide torque for movement.

You can add as many Inline Reaction Wheel modules as is necessary. More of these should allow you to turn easier, maintain a heading, and compensate for unbalanced rockets (within reason), they will also use more electrical power.

Summary:

- ASAS is gone, it has been replaced with SAS

- All command pods, cockpits, and probe cores have SAS, all four parts under the Control Part tab also have SAS

- SAS is less aggressive than ASAS, but more versatile

- Manual input may be required to maintain a course

- It is not an autopilot, but it won't waste RCS like the old ASAS did, and it should make docking far simpler

- SAS is the only thing you are activating/deactivating with the "T" and "F" keys

- Command pod torque has been replaced with the reaction wheel system, this allows for turning independent of any other control system

- All command pods, cockpits, and probe cores have reaction wheel torque to varying degrees

- The Inline Advanced Stabilizer, Inline Reaction Wheel, and 2.5m ASAS module also have reaction wheel torque and SAS functionality

- Torque can be disabled with action groups or through the right-click toggle button

- Torque uses electricity

Troubleshooting:

If you are having issues with maintaining a heading, uncontrollable spinning, or other control problems then there a few steps you may want to take.

Make new crafts from scratch. Start with something simple just to check if everything is working right. Vanamonde and stupid_chris have good examples on page 2 of this thread. Making and testing similar crafts will allow everyone to make sure that we all have consistent flight behavior.

If you use a joystick/gamepad then unplug it and uninstall any drivers (even if you aren't actually using it) to make sure this isn't causing any issues. This still appears to be an issue in 0.22, so it's a good idea to check this if you are having problems. There are several threads in the support forums concerning joystick support.

Be careful to avoid having parts clip through each other. This is known to cause phantom rotational forces; this may not be the only cause of this bug.

Completely delete your KSP folder and make a fresh reinstallation. Make backups if you want, but it's probably best to start everything from scratch; it's still early in the game's development, you should expect to have to do this with every update.

If you find anything wrong with this thread please tell me and I will update this post.

Edited January 14, 2014 by DMagic
Changed title

[Kaleb]

Excellent write-up! Agrees with my findings as well.

Semantically, it's odd, though, that I find myself thinking of the new SAS-functionality in terms being an "ASAS" (because it is a computerized PID-controller with no torque, like the old ASAS) rather than a "SAS" (even though it only kills rotation, like the old SAS); and thinking that it is the old SAS that "went away" (since there is no longer a piece that provides non-user-controllable torque).

The key point that I'm taking away from all of this is that if you expect to be able to steer anything unmanned, you're going to have to at the very least put something with SAS-functionality on it.

[UmbralRaptor]

There is still such a thing as ASAS, even if no stock parts exploit it. It's fairly straightforward to reimpliment, but I'm unsure of how well it plays with the new SAS. Edited July 25, 2013 by UmbralRaptor

[RocketPilot573]

Okay, but how do we dock if our ship is spinning (SAS wont stop it)? How do we fly a plane long distances (SAS wont use control surfaces)? Finally, how in Kerbin do we do a long burn with a ever so slightly asymmetrical craft (SAS doesn't care)?

[NoMrBond]

Seems about right, the new system is two distinct parts

Flight Control [previously ASAS] and Manoeuvre Authority [previously SAS, lin/rotPower and Torque]

Flight control is capable of automatically exercising manoeuvre authority when activated

Command Modules have both flight control and limited manoeuvre authority

ASAS parts just have just flight control

SAS parts, probe cores, canards, winglets and other aero parts just have just manoeuvre authority. Reaction wheel parts require electricCharge to affect the vessel. Aero parts require atmosphere to affect the vessel

If your Flight Control cannot access sufficient Manoeuvre Authority to perform what you're asking your vessel to do, it will feel laggy, unresponsive or otherwise won't comply.

Edited July 25, 2013 by NoMrBond

[Frederf]

Can you explain how the new SAS does its math? "It holds a heading" "It's not an autopilot" how do you reconcile that contradiction? If it holds an attitude it's an attitude autopilot.

What PID values does it use? How does it derive them? When are new attitude target values captured? What control surfaces does it use and when?

[Pingonaut]

Okay, but how do we dock if our ship is spinning (SAS wont stop it)? How do we fly a plane long distances (SAS wont use control surfaces)? Finally, how in Kerbin do we do a long burn with a ever so slightly asymmetrical craft (SAS doesn't care)?

This. Without using MechJeb to keep a certain angle, how are we supposed to do this? Are we now expected to micro-manage things, making processes more boring and annoying, rather than easier and more stable? I don't want an autopilot but I also don't want to do unnecessary tasks to keep my ship straight.

[NoMrBond]

The new system is not a PID controller, Harvester said they would ask c7 to explain the functionality of the 0.21 system

[Vanamonde]

I am baffled. I tried a small rocket and launched one test with each of the guidance pieces, and NONE of them could hold a heading even just coming straight up off of the pad. I tried combining the ASAS with the others, and still nothing. Surely there must be some way to make a ship simply fly straight without constant admonishments from the player?

[stupid_chris]

I had absolutely no problem making rockets hold their headings. Really just look:

Before launch, throttle up, SAS on.

Shortly after launch, everything is straight and nice. Did not touch the controls once.

5km up, still haven't touched the controls once, still haven't moved one inch.

Honestly... the SAS module is doing it's job flawlessly here. I just can't understand how it doesn't work for you guys. If you have to control over your rocket, it should hold it's heading.

[Kaleb]

The new system is not a PID controller, Harvester said they would ask c7 to explain the functionality of the 0.21 system

Interesting... Was not aware it was not a PID. That raises some questions.

[stupid_chris]

Basically, from playing myself, the SAS module works with a combination of the old SAS and ASAS modules together. ASAS was a flight computer that used gimbals, RCS, and other things to hold a heading. SAS was used to stop your rotation using torque. The new SAS uses control wheels, RCS, gimbals and other things to stop your rotation. Like the old SAS, it lets you move around when it's activated, and as soon as you stop, he will try to stop your rotation. That's exactly what the new SAS module is doing.

If you have issues with the SAS modules, I have a few theories:

1) You lack control authority or stability

2) you are using a mod that interferes with the SAS module

3) I'm not sure on that one, but it might be because you transfered a .20 save, but that needs confirmation.

[Vanamonde]

This test ship

has cabin torque, both versions of the inline thingy, canards, and a gimballed engine, but just coming straight up off of the pad with no input from me, this is how quickly and severely it wanders from its heading.

Now, am I misunderstanding something, or are all those control devices combined not supposed to be enough to make this thing fly straight?

[stupid_chris]

Canards are built to go at the front of a ship, they tend to do weird stuff at the back, especially on light ships. Try putting winglets just to see, pretty damn sure it will work.

[stupid_chris]

Okay, I can confirm that the problem is 100% on your game:

Javascript is disabled. View full album

I rebuilt your rocket identically, tried to fly it with both winglets and canards, and it performed perfectly on both tries. I did not touch any button once after pressing space. Whatever the problem is, it's more complex than it looks.

Did you import your save from an old save file?

[Umlüx]

i get the same SAS behavior like Vanamonde

i completely deleted KSP and let steam install the fresh, updated version. new savefile, no mods..

but all my rockets are drifting away from their inital heading. this is quite frustrating.

[Vanamonde]

I downloaded a fresh copy of 21 and built the ship in this version. I often place canards on the low end of my launch vehicles without problems, but even ignoring that issue, I originally mounted the Delta Deluxe Winglets and had the same problem, which is why I replaced them with the larger canards in hopes that they would be stronger, but it didn't help.

[stupid_chris]

I must admit I am beyond confused. Really, I know you from around Vanamonde, and I trust you 100% on your claims, but for the love of Jeb, I mean, I did not fake my pictures neither. I mean I literally built the same thing as you piece for piece, dropped it on the launched pad, throttle up, pressed t and then launched away, and it flies like if it was on rails!

I honestly don't understand at all. There's got to be a common factor to all of you guys who can't get it to work.

EDIT: Wait a second. Umlux, I see you downloaded the game on steam. Vanamonde, do you have a store version or a steam version? Nevermind you have a store version. Just a shot in the dark, but can you confirm that the game version is v0.21.0.272?

Edited July 25, 2013 by stupid_chris

[sumghai]

Might the discrepancies have something to do with varying max physics delta-time per frame across installs?

[FenrirWolf]

I made sure to just throw out my old game entirely and download it fresh. So far I haven't encountered any anomalies with the new SAS and reaction wheels systems, other than that it's not quite as aggressive as the old ones. It's been keeping my stuff on course so far though.

[Danger Will Kerbinson!]

Vanamonde is not alone. I didn't even mess around trying to save anything, just wiped all local files from Steam, ditched the entire KSP folder inside of steamapps and let it all come in fresh.

I have had rockets veer off course right at launch and then continue that new unplanned heading very well like it was just a momentary lapse. Sometimes they fly perfectly and respond to inputs I give, other times they decide to put some other directions into mix for fun. For example during a gravity turn it should just be a matter of a few pushes of D to make the turn right on the 90 degree line and all other axes are to remain locked, but they are not...sometimes it drifts up and down at random, with the simplest of ships.

Other problems include the inability to hold to a maneuver node marker during a warp. I don't mean how sometime you would hit warp and the blue mark would jump a half inch or so, but it will drift during the warp when things are not even supposed to be subject to physics, its just a nice slow drift while at 50x warp.

When I set the ship to face Normal to watch where it would drift and how fast it won't do it. Stays on that mark for rotation after rotation like it should with that heading. But try to get it to hold any other heading and it just drifts and sways, thrust, no thrust...doesn't matter.

Clearly the program is not behaving the same for everyone.

[Danger Will Kerbinson!]

Might the discrepancies have something to do with varying max physics delta-time per frame across installs?

I guess it could be a lot of things, two that spring to mind are differences in computers, with perhaps faster rigs like mine allowing shenanigans to get through OR perhaps it is Steam users who have a bug in the update.

[stupid_chris]

Did you guys try to delete your settings file and reboot the game? Harv proposed that earlier today, just wondering if that worked or not.

[sojourner]

I said this in another thread, but could it be platform related? Who's running Windows vs. IOS vs. Linux?

I personally run Windows and am having none of the issues others are having with SAS.

[stupid_chris]

Vanamonde runs Windows and he has issues. Windows here to and everything works.

EDIT: Guys, random question, but how many of you who have problems are using a joystick?

Edited July 25, 2013 by stupid_chris