3.75 m parts + SAS = Death Wobble

[PTNLemay]

At first I thought it might be something to do with the new aerodynamics, but I'm seeing it even in space. I cut everything, time accelerate to nullify any little rotations that might be present, return to normal speed, turn on SAS, then increase the throttle. If it goes above 60% or so the SAS starts overcompensating imediately and the whole rocket starts to wobble aggressively. I have to quickly turn off SAS and guide it manually otherwise the rocket rips itself to pieces.

I don't remember it does this prior to 0.90, though... maybe I'm wrong, I rarely used the giant parts. Am I doing something obviously wrong here? I've seen people use 3.75 m parts on streams and things like that, and they don't seem to rip themselves to pieces like mine are doing.

http://i.imgur.com/Uv0TL55.png No SAS

http://i.imgur.com/HGEU3xJ.png With SAS, you can see the kink in the middle where the stack separator meets the 2.5 m SAS module. It sways wildly back and forth, each pulse getting bigger and bigger.

[Randazzo]

At first I thought it might be something to do with the new aerodynamics, but I'm seeing it even in space. I cut everything, time accelerate to nullify any little rotations that might be present, return to normal speed, turn on SAS, then increase the throttle. If it goes above 60% or so the SAS starts overcompensating imediately and the whole rocket starts to wobble aggressively. I have to quickly turn off SAS and guide it manually otherwise the rocket rips itself to pieces.

I don't remember it does this prior to 0.90, though... maybe I'm wrong, I rarely used the giant parts. Am I doing something obviously wrong here? I've seen people use 3.75 m parts on streams and things like that, and they don't seem to rip themselves to pieces like mine are doing.

http://i.imgur.com/Uv0TL55.png No SAS

http://i.imgur.com/HGEU3xJ.png With SAS, you can see the kink in the middle where the stack separator meets the 2.5 m SAS module. It sways wildly back and forth, each pulse getting bigger and bigger.

Lock the gimbal on the engine and let the reaction wheels do their thing. It will still wobble a little, but tends to be less likely to snap in two.

You see, what happened was, Squad decided that the 0.90 SAS was too useful. In an effort to up the difficulty a touch, they kidnapped a cab driver in Mexico, got him drunk, then hooked him up to a Kermancephalogram machine and cloned his brain waves while driving. Thus, 1.0x SAS was born.

Edited June 10, 2015 by Randazzo

[PTNLemay]

@Randa

Hmm, I'll try that, thanks. I've tried turning the reaction wheels off, but I didn't think of the gimbals.

[gogozerg]

Definitely a problem with SAS.

You may have a stable rocket flying straight, and when you activate SAS (a stability augmentation system) it starts to wobble.

People here will tell you that your rocket has not enough struts to make it a brick, or that you should lock the gimbals.

[Jouni]

Your rocket has too much control authority (and too many joints and a too powerful engine). 1.0 increased the gimbal range of most engines by 2-3x, so you often have to tweak it down to make the rocket less wobbly.

[PTNLemay]

Yep, removing the engine gimbaling helps loads.

@ Jouni

But you can't tweak it down, can you? From what I can tell you can only toggle it on or off, and indeed when it's on it's way too much range of motion.

[mythbusters844]

Locking gimbal does help, but I'm pretty sure that wobbly joints are a unity issue and can't be fixed. :^)

Just kidding, Ferram made a mod to make joints stiffer and stronger if you're okay with installing mods.

[FonVegen]

@PTNLemay Actually there is a bar for "Gimbal Range" that you can in- or decrease from 0 to 100.

[LordFerret]

I've had problems with SAS fighting itself since I first started playing (v0.23.5), and the same problems with it in v1.0.x ... only more pronounced now. These days I skip the SAS modules altogether and just go with RCS; Also, struts (no need for mods). That way I don't have to worry about tweaking gimbals.

[JoCRaM]

Those large reaction wheels always used to be really flimsy.

[TrooperCooper]

Struts and RCS drive up your part count. Struts also add drag and make your rocket look ugly as...

With the Joint Reinforcement mod you can fly huge vessels without struts, without wobble... guess what... instead of looking like pyramids or pieces of strut-art, your creations will automaticly begin to look like... rockets.

[LordFerret]

...except I prefer to play with a minimum of mods (stock bug fixes, scansat, lazor docking cam - that's it).

[mikeb30165]

In a nutshell, switch-off gimballing, stiffen up your design and move the SAS as close to the CG as possible. By moving the SAS closer to the CG you're minimising bending movements, and to stiffen, try strutting between the stages and (this might sound daft) strut the SAS unit/s to the rockets; I've noticed the joints between the units and the rest of the ship get a little bendy on big ships (especially if placed at the ends), causing oscillations that the SAS's overcompensate for, further exacerbating the problem. Also, your design uses the big Rhino engine, which is massively over gimballed (IMHO), so switch the gimbals off on it (as others have suggested). Just my tuppence worth, & hope it helps..

[Netskimmer]

...except I prefer to play with a minimum of mods (stock bug fixes, scansat, lazor docking cam - that's it).

I tend to think of wobbly rockets and a bug/design issue (I can't imagine that Squad intended them to act like that). So for me Structural Reinforcement Mod would fall under the category of stock bug fix.

[PDCWolf]

Please let's not criticize wobble, remember it's there to be fun and Kerbal. Now, as a matter of fact, wobble is fixable (so are many other issues), fully, but it has been said by squad many times that they want to keep it for the supposed challenge, fun and kerbalness it gives.

Just download Kerbal Joint Reinforcement and, to help your many mod issues, just pretend it's one of those "stock fixes" (it actually is).

[bakanando]

The rockomax 2.5 decoupler is also a very wobbly part, after switching to separators of the same size I've had little to no wobbly rockets in past versions. I reduce the use of the decoupler as much as I can for that reason.

[selfish_meme]

The rockomax 2.5 decoupler is also a very wobbly part, after switching to separators of the same size I've had little to no wobbly rockets in past versions. I reduce the use of the decoupler as much as I can for that reason.

Makes sense, it's for decoupling things, while the separator is for stacking stages. Don't mind we did not have a seperator for a long time.

[Jouni]

I tend to think of wobbly rockets and a bug/design issue (I can't imagine that Squad intended them to act like that). So for me Structural Reinforcement Mod would fall under the category of stock bug fix.

The amount of wobble was a deliberate choice. It used to be much higher, but Squad reduced it in 0.23.5, when they introduced 3.75 m parts. It would have been trivial to eliminate it completely (at least for all even remotely sane designs), but they didn't.

But, as you say it, wobbly rockets are a design issue. Badly designed rockets wobble, while well-designed ones usually won't. These days you need one strut per booster (to create a two-point connection) and a few struts around undersized connection nodes, but that's it. If you need more, it's usually a sign that your rocket design is bad.

[Randazzo]

But, as you say it, wobbly rockets are a design issue. Badly designed rockets wobble, while well-designed ones usually won't.

SAS can introduce wobble in the best designed of rockets, and can and will tear large ones apart, regardless of the ego of the builder.

[Jouni]

SAS can introduce wobble in the best designed of rockets, and can and will tear large ones apart, regardless of the ego of the builder.

If SAS introduces significant wobble, the rocket is by definition badly designed.

In my experience, the primary reason for wobbly rockets is that aesthetic considerations have taken priority over engineering considerations. There may be too many joints, too powerful engines, or too much control authority. The mass of the rocket may be badly distributed, torque/drag/lift may be applied at wrong positions, or the struts may be badly placed. Whathever the reason is, it's an engineering problem that can usually be solved with a better design.

Wobble may be a more serious problem, if you use mod parts larger than 3.75 m. But if you restrict yourself to stock parts and to rockets that are no taller than the internal dimensions of the VAB, the problem is almost always in the design, not in the joints.

[Randazzo]

If SAS introduces significant wobble, the rocket is by definition badly designed.

In my experience, the primary reason for wobbly rockets is that aesthetic considerations have taken priority over engineering considerations. There may be too many joints, too powerful engines, or too much control authority. The mass of the rocket may be badly distributed, torque/drag/lift may be applied at wrong positions, or the struts may be badly placed. Whathever the reason is, it's an engineering problem that can usually be solved with a better design.

Wobble may be a more serious problem, if you use mod parts larger than 3.75 m. But if you restrict yourself to stock parts and to rockets that are no taller than the internal dimensions of the VAB, the problem is almost always in the design, not in the joints.

Let me try a different approach. It is a known and undeniable fact that the SAS overcorrects badly, and the increased engine gimbal range has made the problem worse.

You are correct, it is possible by applying otherwise unnecessary limitations to your vehicles, to design around the problem. This does not mean there is no problem, or that vehicles not designed specifically to get around it are bad design.

For example, a vehicle that snaps in half with gimbals unlocked that flies a perfect ascent with them locked is not a poorly designed vehicle.

Edited June 11, 2015 by Randazzo

[rkman]

Incorrect, again. When locking engine gimbal results in a perfect ascent, the problem does not lie within the rocket design.

It can be argued that locking engine gimbal (or otherwise reducing control authority) is an adjustment of the rocket's design.

It's not a secret that SAS badly overcorrects. It can't be denied away.

That is for sure, but then still a rocket can have to much of something (even boosters or struts) even though a properly tuned SAS might be able to deal with it.

If anything, the game is unhelpful with finding the cause of all those disintegrating newbie rockets.

[Jouni]

Let me try a different approach. It is a known and undeniable fact that the SAS overcorrects badly, and the increased engine gimbal range has made the problem worse.

SAS does not overcorrect badly, except for individual designs. It's a PID controller with a fixed set of parameters. Because we can't adjust the parameters in the stock game, we have to adjust the amount of control authority instead.

Even when the SAS does overcorrect, it just causes oscillation, not wobble. To turn that oscillation into wobble, the rocket has to be structurally unsound. In atmospheric situations, poor piloting can also cause wobble.

[tutike2000]

This is why I finally gave up and started to use MechJeb. I hate SAS overcompensating every damn time.

SmartA.S.S. tends to do it better, but if you go past 300-ish parts even mechjeb can't stop the wobble.

[Randazzo]

SAS does not overcorrect badly, except for individual designs. It's a PID controller with a fixed set of parameters. Because we can't adjust the parameters in the stock game, we have to adjust the amount of control authority instead.

Even when the SAS does overcorrect, it just causes oscillation, not wobble. To turn that oscillation into wobble, the rocket has to be structurally unsound. In atmospheric situations, poor piloting can also cause wobble.

It does, in fact. No matter how many times you declare it doesn't, it continues to function poorly. It can be tested and seen with a command pod in orbit. There's really no point to arguing it with you further since it can be easily demonstrated without any accompanying "design" to influence it's behavior.