SAS is literally useless

[FlazeTheDragon]

Like, i get that this is not optimal reaction control placement, but properly tuned SAS P.I.D's should be able to easily cope with this.

Current SAS is borderline unusable on most builds.

 

[Razor235]

Devs have said the PIDs still need tuning, but it really doesn't look like it even HAS Proportional control, let alone the other two.

[pandaman]

I'd say it's just 'different' to what you may be familiar with from KSP1.  Try using more stabilisation and/or RCS.

No doubt they still have tweaking and tuning to do too.

 

[FlazeTheDragon]

8 minutes ago, Razor235 said:

Devs have said the PIDs still need tuning, but it really doesn't look like it even HAS Proportional control, let alone the other two.

Ive been flying FPV drones for a fair number of years and what im seeing honestly just looks like extremely poorly tuned PID's. I tried to go in to the config files and see if i can tune them me self, but they use some sort of an auto-tune algorithm which which i sadly couldnt figure out.

My guess is that Integral is the main culprit, bumping it up would for sure make things a lot better, tho im sure its not as easy as that and more fine-tuning needs to be done.

3 minutes ago, pandaman said:

I'd say it's just 'different' to what you may be familiar with from KSP1.  Try using more stabilisation and/or RCS.

No doubt they still have tweaking and tuning to do too.

 

Its deffo very different from KSP1, considering SAS was actually useful there lol.

Adding RCS doesnt really change anything since RCS is still controlled by the same PID's, so you get the same result. And reaction wheels are impossible to use on ssto's since they have bugged drag calculations. If i added a single reaction wheel to that ssto the drag would double. + those reaction wheels would still be controlled by the same poorly ''tuned'' SAS.

The only real solution for more complex builds is to turn off SAS completely and just fly everything manually with trim, which is extremely tedious.

Edited March 14, 2023 by FlazeTheDragon

[shdwlrd]

Seen this behavior before in KSP1. The torque is being applied at a different angle than expected. It's simply the model or torque direction is wrong. To counter act this, turn off your reaction wheels in the pods and cores and don't use the inline stabilizers. Just use RCS and aerodynamics. This has fixed the SAS issues for me.

Edited March 14, 2023 by shdwlrd

[KerikBalm]

So it's not SAS that is broken, but reaction wheels?

[MechBFP]

7 minutes ago, shdwlrd said:

Seen this behavior before in KSP1. The torque is being applied at a different angle than expected. It's simply the model or torque direction is wrong. To counter act this, turn off your reaction wheels in the pods and cores and don't use the inline stabilizers. Just use RCS and aerodynamics. This has fixed the SAS issues for me.

I strangely haven’t had any issues with SAS in KSP2. It works perfectly fine for me (other than flappy bird planes at high speed), although now that I think of it I don’t think I used any reaction wheels in any of the builds aside from the one built into the MK3 command pod and the MK1 cockpit. 

Edited March 14, 2023 by MechBFP

[TackleMcClean]

Excuse rookie question - what are "PID"s?

[FlazeTheDragon]

18 minutes ago, shdwlrd said:

Seen this behavior before in KSP1. The torque is being applied at a different angle than expected. It's simply the model or torque direction is wrong. To counter act this, turn off your reaction wheels in the pods and cores and don't use the inline stabilizers. Just use RCS and aerodynamics. This has fixed the SAS issues for me.

Only using RCS is a decent work around as long as you have fairly low rcs authority. Building in stronger RCS authority gives fairly similar results, not to mention that sas will waste a lot of rcs propellant trying to keep you steady since its very inefficient. That said, having to build in a bunch of thrusters in to an ssto is pretty rough since rcs thrusters have (i think) more drag then they should.

 

16 minutes ago, KerikBalm said:

So it's not SAS that is broken, but reaction wheels?

Nah, anything that is able to be controlled by SAS is busted, rcs, reaction control wheels and cockpits all use the same PID's. I only got that 1 inline cockpit in the ship from the vid.

4 minutes ago, TackleMcClean said:

Excuse rookie question - what are "PID"s?

PID's are basically a fly-by wire algorithm that is designed for craft control, like planes, rockets, quadcopters etc. Its basically just a program that is designed to keep the craft stable by knowing when to automatically apply reaction control in any of the 3 (x y z) axis.

Edited March 14, 2023 by FlazeTheDragon

[Razor235]

5 minutes ago, TackleMcClean said:

Excuse rookie question - what are "PID"s?

Stands for Proportional Integral Derivative controllers.

 

Proportional (position only):  The further away I am from my target, the more I push towards it.

Integral (position integrated over time):  I look at my past data to see how far I've moved over some time relative to where I am now, so I can adjust to smooth and prevent deviation.

Derivative (position with respect to time):  I look at how fast I am going, so I reduce overshooting at the cost of increasing errors because I don't see where I actually am.

 

Properly balancing the three will get you a very smooth, predictable, and responsive behavior.

Too much P will cause responses too fast and large for the others to compensate.

Too much I will cause sluggish response and may not reach the actual steadystate target.

Too much D will cause excessive wiggling around the steadystate target.

 

Edited March 14, 2023 by Razor235

[Wheehaw Kerman]

The reaction wheels in KSP1 were hugely OP and extremely unrealistic.  No saturation, just for starters.  One of the reasons I liked the Soviet parts from Making History was that they loaned themselves well to more realistic RCS-only builds.

KSP2’s nerfing the reaction wheels is an improvement IMHO.

[MechBFP]

42 minutes ago, Wheehaw Kerman said:

KSP2’s nerfing the reaction wheels is an improvement IMHO.

They aren’t actually nerfed but aren’t working as intended currently it seems like because of the intertial tensor scalar. 

[Wheehaw Kerman]

11 minutes ago, MechBFP said:

They aren’t actually nerfed but aren’t working as intended currently it seems like because of the intertial tensor scalar. 

Hopefully they do actually nerf them a bit.  The reaction wheels were bit of an immersion breaker.  Every time I used them I had this nagging thought that I shouldn’t be able to convert electricity to angular moment that easily :).

[cocoscacao]

Cockpits also have OP reaction wheels. Hope they get nerfed too. 

[Scarecrow71]

2 hours ago, FlazeTheDragon said:

Adding RCS doesnt really change anything since RCS is still controlled by the same PID's, so you get the same result. And reaction wheels are impossible to use on ssto's since they have bugged drag calculations. If i added a single reaction wheel to that ssto the drag would double. + those reaction wheels would still be controlled by the same poorly ''tuned'' SAS.

Unfortunately, the answer really is to spam RCS.  Not just a couple of thrusters, but like "moar RCS".  You can also try to mess with the engine gimbal and turn that down a bit for a bit more control.

[FlazeTheDragon]

5 minutes ago, Scarecrow71 said:

Unfortunately, the answer really is to spam RCS.  Not just a couple of thrusters, but like "moar RCS".  You can also try to mess with the engine gimbal and turn that down a bit for a bit more control.

Having powerful RCS leads to the exact same issue since RCS is controlled by the same very poorly tuned PID's.

@MechBFP @Wheehaw Kerman @cocoscacao If you guys want the game to be more realistic nothing prevents you from simply turning reaction controls off. I feel like needing rcs on every craft would make ksp even harder on new players so its a good system to have for accessibility sake.

[Wheehaw Kerman]

4 minutes ago, FlazeTheDragon said:

Having powerful RCS leads to the exact same issue since RCS is controlled by the same very poorly tuned PID's.

@MechBFP @Wheehaw Kerman @cocoscacao If you guys want the game to be more realistic nothing prevents you from simply turning reaction controls off. I feel like needing rcs on every craft would make ksp even harder on new players so it’s a good system to have for accessibility sake.

I can’t disagree.  The reaction wheels are a not-terrible easy-mode crutch for new players.  I’d be fine with not nerfing them all the way to real-world levels.

ETA: wow I came across as arrogant there.  Basically, yeah, you’re right.

Edited March 14, 2023 by Wheehaw Kerman

[Scarecrow71]

7 minutes ago, FlazeTheDragon said:

Having powerful RCS leads to the exact same issue since RCS is controlled by the same very poorly tuned PID's.

@MechBFP @Wheehaw Kerman @cocoscacao If you guys want the game to be more realistic nothing prevents you from simply turning reaction controls off. I feel like needing rcs on every craft would make ksp even harder on new players so its a good system to have for accessibility sake.

I'm using a combination of RCS and reaction wheels to control craft in orbit (and in space, in a general sense), and I'm not having issues when doing that.  It takes a lot of RCS thrusters and a lot of monopropellant to deal with it, but it's working for me.  I'd ask if you've actually tried using it, but I assume you have?

[cocoscacao]

@FlazeTheDragon true, but manually setting everything bores me greatly. Maybe they could add that in global difficulty settings. 

[FlazeTheDragon]

5 minutes ago, Scarecrow71 said:

I'm using a combination of RCS and reaction wheels to control craft in orbit (and in space, in a general sense), and I'm not having issues when doing that.  It takes a lot of RCS thrusters and a lot of monopropellant to deal with it, but it's working for me.  I'd ask if you've actually tried using it, but I assume you have?

yup, i put rcs on most of my bigger craft that actually need to maneuver outside of chilling in orbit. Id suggest using verner rcs, the small monoprop blocks are honestly only useful for docking.

[Jacob Kerman]

38 minutes ago, Wheehaw Kerman said:

Hopefully they do actually nerf them a bit.  The reaction wheels were bit of an immersion breaker.  Every time I used them I had this nagging thought that I shouldn’t be able to convert electricity to angular moment that easily :).

Honestly, I'm inclined to agree. Turning a 200-ton ship in orbit in ksp's physics just by using reaction wheels was pretty weird. Made RCS useless on upper stages and landers.

[regex]

The biggest problem with falling back on RCS in KSP is that it isn't precise enough to hold a heading without flopping around wasting fuel, or it doesn't use just enough fuel to start a turn before waiting until aligned to use just enough more fuel.  In short, it's not smart. Reaction wheels with decent control authority are absolutely necessary to avoid these issues without relying on an experienced user to handle and remember all this stuff. I don't mind the strength of reaction wheels in KSP2, they seem toned-down from KSP1 (which was ridiculous), although it would be nice to require desaturating them with RCS on occasion, I think that would be a good compromise.

E: Plus, I need decent reaction wheels to flip my rover over when the terrain randomly throws it into the air.

Edited March 14, 2023 by regex