Are wobbly rockets back?

[Wintersdark]

Anyone who's worked with FAR (or NEAR) will recognize the "wobbly rocket" aspect of the streams to be that the streamers are generally launching lawn darts with almost no steering capabilities and without anywhere near the correct profile for that to work out well. If they had stabilization off, they'd go even more off-course. Fins are your friend in the new atmosphere! So is not pointing a billion miles away from prograde while in an atmosphere thicker than 0.02atm.

This.

At least, all the streamers I saw before irritation overrode curiosity fell victim to this.

Large, long rockets typically with a narrow (read: weak) section in the middle and a total lack of fins, despite everyone in the chat room screaming FINS FINS FINS. Generally, a very large amount of SAS as well. I suspect this is due to habit more than anything else, in .9 fins where totally unnecessary and SAS was pretty damn OP; simply stacking on more would magically stabilize anything.

- - - Updated - - -

We never needed tutorials.

Some people don't. Some already have a grasp of how this works, others like spending hours on the forums/watching videos. Still some more like to just struggle-till-it-works, Dark Souls style. Those are all perfectly valid and awesome ways to go. Me, I'm a struggle-till-it-works+reading forums guy.

But there are a LOT of people out there who learn better by being taught. That's not wrong either. They've been largely excluded from KSP thus far, and I'm sure Squad would like to bring more people into the fold. I know a great many people who dropped the game after several catastrophic failures. In most cases, where they didn't understand why they failed, and thus didn't know how to fix it.

Tutorials are a good way to go here. At least for The Basics. Education is a great thing.

Except... Please, not like the current crop of mobile games, that insist on telling you every single click to make in unskipable tutorials. Gah.

[John FX]

Yes. The wobble is induced by the SAS over-correctting with gimbal. But SAS and gimbal are not the cause and are not to blame. It's like blaming ABS, EPS and other systems for crashing your car while doing 200kph on ice. They are simply not designed to cope with those conditions. If you want the system to fly your rocket for you wait a few days and install MechJeb. If you want to fly your rocket yourself YOU have to fly it. and YOU have to fly it properly.

Um, no. Using SAS is not 'getting the system to fly your rocket for you' and using flight aids is not a black and white digital choice where either you control literally every aspect of the rocket without using SAS or you go full autopilot.

I reject your false dichotomy.

SAS should cope with the new aero. It should be designed to cope with the conditions encountered in flight. It should not overcorrect. Even mechjeb suffers from this and there are plugins to allow it to work well with FAR.

[Tex_NL]

Exactly. SAS is an AID. And if you use it properly it WILL do what it's supposed to do.

[Superfluous J]

The problem is we've seen very few instances of people using it properly.

[Laie]

You just contradicted yourself. You said to start the gravity turn at 25km, and then said to start it as early as possible.

Key message that didn't come across: you should start the turn early, not do a 30-degree turn it all at once.

Gravity Turn 101:

Build a very simple rocket, but with tail fins[/]. Use launch clamps and tilt the whole assembly slightly (Shift-D or use the rotate gizmo). Put it on the pad, launch, and don't touch the controls: Just see what happens. Try that a few more times with different TWRs (reasonable values for this purpose go from 1.5 .. 3.0).

Even in stock atmo, this works good enough to illustrate the concept. Three to five minutes of messing around may teach you more about gravity turns than many a learned forum post.

[Nobody_1707]

True! ÃŽâ€V and TWR should be displayed stock. Always visible in VAB/SPH and through the skills of one of your Kerbal engineers during flight. But what good is a tool without the skills to wield it? All of this is useless without proper tutorials.

I was shocked that the TWR & ÃŽâ€V weren't in the Engineering Report of the VAB, I knew that the in flight readout for those was delayed, but the Engineering Report was the perfect place to put them while designing the rocket.

[r_rolo1]

Yes. The wobble is induced by the SAS over-correctting with gimbal. But SAS and gimbal are not the cause and are not to blame. It's like blaming ABS, EPS and other systems for crashing your car while doing 200kph on ice. They are simply not designed to cope with those conditions. If you want the system to fly your rocket for you wait a few days and install MechJeb. If you want to fly your rocket yourself YOU have to fly it. and YOU have to fly it properly.

Well, I disagree with you in the part of the SAS being not to blame. Sure, the primary blame is on the rocket designer/pilot , but the fact that the SAS increases instability is something that is probably not a thing we should forgive because the pilot is a idiot, in the same way we don't blame a malfunctioning ABS system on the driver , in spite of being him that has put the car in a wheel lock situation.

Remember that SAS stands for stability augmentation system ... if a system that is supposed to increase your stability is actually decreasing it, it is not exactly working as expected That does not mean that it should be a Mary Sue savior, but atleast it should behave like it says on the tin and not exactly in a opposite way

Edited April 26, 2015 by r_rolo1

[Fel]

Well, I disagree with you in the part of the SAS being not to blame. Sure, the primary blame is on the rocket designer/pilot , but the fact that the SAS increases instability is something that is probably not a thing we should forgive because the pilot is a idiot, in the same way we don't blame a malfunctioning ABS system on the driver , in spite of being him that has put the car in a wheel lock situation.

Remember that SAS stands for stability augmentation system ... if a system that is supposed to increase your stability is actually decreasing it, it is not exactly working as expected That does not mean that it should be a Mary Sue savior, but atleast it should behave like it says on the tin and not exactly in a opposite way

There have been LONG posts in the past on SAS and how it works "in real life", to summarize: a SAS system is tuned differently for EACH individual craft that uses it. Of course, a sas sysem is likely to need multiple reference points to work correctly but I don't know how KSP's system works.

KSP also has problems since it lacks analog input... or well, any professional sense of controls. Tilting the gimbal to 12*N 35*W (it should have 2 axises of movement) is impossible without mods; and I don't know if there ARE any mods that supply an easy access level of control like that. (well, it possibly could be tapered, but not individually)

It should be possible to throw RCS units all over the ship at random and balance the system using mathematics (a level that squad doesn't employ) (i.e. by controlling how much thrust is applied on one end, or using a 'digital' effect of "on off on off on off" to equate an equivalent level of thrust, various thrusters can be fired to counteract the poorly balanced placement)

Edited April 26, 2015 by Fel

[NathanKell]

It's not that wobble is back. It's that it's always been present, but being hidden by the fact that aero didn't use to actually be aero. People who've flown with FAR know that wetnoodlism has never been gone though .23.5 did help a lil' bit (but nerfed struts).

And yes, it'd be nice if the streamers actually took the time and effort to figure out how new aero worked before lolsplosions.

[r_rolo1]

@Fel

While I agree with you in the general sense, I think that the issue behind the SAS overcorrections that are magnifying the bad piloting we're seeing in the streams is created by a badly designed correction output curve. In other words, the game at points gives too much of a response to the deviation ( and in others it simply refuses to correct, like when your control authority is below 200% of what you need ... that is a issue that plagues 0.90 spaceplanes , BTW ). While a finely tuned SAS would need to be custom made per ship, IMHO the current system is still quite coarse compared with what it could be without having to overly lose generality ... and it is not that SAS is some sacred cow of SQUAD, since they already changed it once ( to better, I might add )

Edited April 26, 2015 by r_rolo1

[Tex_NL]

Let's compare the stock SAS to MechJeb's Smart A.S.S. to demonstrate what SAS does and how it should be.

When you ask the stock SAS to rotate a craft in orbit from prograde to retrograde you'll notice it will overshoot it's target each and every time. The RCS and reaction wheels keep rotating the craft until it hit the target, only then will it try to stop the rotation and hold. MechJeb's Smart A.S.S. does the same but instead of overshooting the target it starts to slow the rotation well before the target. It stops dead centre on the target each and every time.

The same thing is happening during launch. The SAS gimbals hard to keep the nose steady but then overshoots and then has to correct even harder. This combined with the new atmosphere and poor piloting results in some very nasty oscillations.

[Renegrade]

It's not that wobble is back. It's that it's always been present, but being hidden by the fact that aero didn't use to actually be aero. People who've flown with FAR know that wetnoodlism has never been gone though .23.5 did help a lil' bit (but nerfed struts).

Exactly! (I'm saying this a lot today :S )

ARM also didn't really make joints that much stiffer, but rather insanely strong (for instance, making a rocket out of 200 round8s in the ARM update--it'll flop around like crazy but generally hold together until it hits the ground)...

And yes, it'd be nice if the streamers actually took the time and effort to figure out how new aero worked before lolsplosions.

Exa.. er, Indeed!

#lolsokerbal~

[Fel]

Let's compare the stock SAS to MechJeb's Smart A.S.S. to demonstrate what SAS does and how it should be.

When you ask the stock SAS to rotate a craft in orbit from prograde to retrograde you'll notice it will overshoot it's target each and every time. The RCS and reaction wheels keep rotating the craft until it hit the target, only then will it try to stop the rotation and hold. MechJeb's Smart A.S.S. does the same but instead of overshooting the target it starts to slow the rotation well before the target. It stops dead centre on the target each and every time.

The same thing is happening during launch. The SAS gimbals hard to keep the nose steady but then overshoots and then has to correct even harder. This combined with the new atmosphere and poor piloting results in some very nasty oscillations.

Having a high level of dampening isn't always desirable, it reduces the speed at which a system reacts as well as how fast it settles within bounds, to the new level. As it stands, KSP is a fair bit underdamped. MechJeb, from your description, has a fixed level of dampening that is not always the best value for the ship at hand, highly unstable ships would BENEFIT from having less dampening than the MechJeb default, poorly constructed ships would benefit from having more dampening.

Either case, we end up with needing tuning.

@r_rolo1:

What I was trying to say is that even if you design a rocket horribly, you should still be able to stabalize it.

Now, whether we accept the "MechJeb" way of reading universe data and allowing the most horribly designed rocket to actually fly (I don't know how far it goes, but, especially if you take control of individual engine thrust and vectoring, yeah, you should be able to do it); Accept the notion of tuning it entirely ourselves; accept poorly tuned components; or find some medium is more so the question.

Edited April 26, 2015 by Fel

[gearbolts]

I had to make an account just to say there are a few people in this thread up their own butts with what basically amounts to "git gud". They're not helping, they're just being elitist and saying the way they do things is so great and that other players are idiots.

[rkman]

What you're seeing isn't poor physics, it's poor rocket design and poor ascent trajectories.

Those are major causes of wobble, but it does not help that SAS is over-enthusiastic nor that engine gimbal movement is instantaneous, nor that max gimbal range has been increased in 1.0.

The workaround usually is to not have to much reaction wheels and engine gimbal, and to not use steerable fins/winglets.

[Laie]

The wobbly rocket problem seemed to be fixed in 0.23.5 and I havn't had any problems since. However in every stream and video of 1.0, even the simplest rockets wobble around like a wet noodle. Has the joint strength been purposefully turned down or something?

If anything, joints have been improved. However...

A) the new aero encourages people to build tall columns, those would have been problematic in 0.9 as well

engine giballing has been much increased

C) SAS hasn't learned to cope with it's new powers

There is a new gimbal strength slider, but it doesn't seem to have any effect. I often have to turn it off entirely.

[WalkingShadow89]

They increased the gimbal range on all of the engines by a ton so that could be it

[Francois424]

I did notice a return of that. Nothing a good strutting won't fix, just like early game.

Also it really doesn't matter if it's under a faring or not, it still wobbles.

And yes, turning off SAS fixes it.

Feels just like the old game to me, heh.

[rkman]

No it isn't. It's people not knowing how to handle the new aerodynamics.

Does not explain why wobble goes away when people turn SAS off.

SAS also does not know how to handle the new aero (and it does not know how to handle rockets of different sizes anyway).

Instant gimbal deflection also does not help. Maybe Squad is saving "useGimbalResponseSpeed" for the properly completed version of KSP?

Otoh, in 1.0.x there sure is more that people can do wrong: to many reaction wheels, to much gimbal, going to fast, turning to fast.

But human error is not the only factor here. It's not like we don't know that KSP in its current state is far from flawless.

[Two]

SAS is definitely broken, but not only in aero. I just landed a rocket retrograde, and even while still in space SAS constantly flickered and wasted huge amounts of electricity for unnecessary auto-corrections. As this rocket (pod + 180L fuel tank + terrier) is unable to wobble for physics reasons (and SAS steady doesn't wobble at all), it just has to be the SAS.

There are however rocket designs that cause wobbling effects, usually caused by decouplers, because those are actually physic joints and not stiff attachments. Those can however usually be fixed with struts.