Roll instability plane issues

[baldamundo]

I don't normally have a lot of problems with spaceplane stability, but trying to work out what's wrong with my latest design is driving me insane.

In level flight (with or without SAS), it's absolutely fine and pitching up or down doesn't cause any problems whatsoever. The issue is solely on the roll axis.

http://imgur.com/gallery/02GHy/new (if someone wants to tell me how to do those fancy imgur embedding things, please do explain)

If you try to roll, it works at first, but it's like it doesn't want to be rolled, and will start trying to roll in the opposite direction. Sometimes this reaction is so strong that it'll not only stop you from rolling in the direction you want to, but you'll actually struggle to stop yourself from rolling perpetually in the opposite direction instead! It's as if the planes roll controls actually make worse the problem they're supposed to fix.

And it's definitely to do with the control surfaces. I only managed to get some stable flight by restricting the control surfaces so each set only controls one axis - e.g. the rudder for yaw, wings for pitch, fins for roll. If you enable roll controls on the wings' control surfaces, the problem gets massively, massively worse. Not only will there be a sort of bounce in the opposite of the direction you're trying to roll in, but you'll not be able to roll at all - e.g. hold down E, for a moment nothing will happen, and then you'll just slowly start spinning in the opposite direction.

The problem's so severe, that if you're in stable flight with SAS enabled, and click to enable roll control on the wings, you'll very quickly find yourself spiralling uncontrollably, and the only way to bring it under control is actually to disable all roll control altogether. It's as if the plane design is naturally stabilising, but actually the roll controls are somehow dangerously counter-productive.

What am I doing wrong?

- - - Updated - - -

I should say, I'm not using any aerodynamic mods like FAR, but the wings and wing control surfaces are from B9 Procedural Parts, the fuel tank is from Procedural Parts, and the cockpit, air intake and engine are from SXT.

EDIT: Additional information - the plane is perfectly symmetrical, the centre of mass (dry and wet) is just behind the cockpit, and the centre of lift just slightly behind that.

EDIT: Huh. It looks like something really screwy is going on with the orientation of the control surfaces.

Edited October 6, 2015 by baldamundo

[Randazzo]

I tried to replicate your design as best I could, and I could only get the described behavior by having roll control only on the rudder itself. My only guess is that the wing control surfaces are somehow inverted.

I think this is my new favorite plane.

Edited October 6, 2015 by Randazzo

[baldamundo]

So after a couple of hours of troubleshooting, I've worked out it seems to actually be a bug with the way KSP calculates a part's position relative to the centre of mass, causing some of my control surfaces to move in opposite directions at times. Possibly it's a procedural parts issue, but it might just be that KSP gets confused by having control surfaces on swept wings near the centre of mass http://imgur.com/gallery/02GHy

- - - Updated - - -

Okay, weirdest thing - the control surfaces near the chassis had their roll axis inverted but the pitch working normally, while the ones near the wingtips had their roll axis fine but the pitch axis inverted ???

EDIT: Wrong link, should have been http://imgur.com/gallery/hwkWT

Edited October 6, 2015 by baldamundo

[SalehRam]

It is better to set each part of the control surfaces to a single task.

I usually set two the control surfaces on the main wings for roll only, the rest are just fixed control surfaces (and back-up in case I need to get more control in mid-flight).

The 2 back tail fins are only for pitching, and the rudder is yaw only.

This way I get some stable control over the plane, and in mid-flight if I feel pitching is slow or roll is slow, I can assign more control surfaces from the main wings to give more control for the lacking part.

[baldamundo]

It is better to set each part of the control surfaces to a single task.

I usually set two the control surfaces on the main wings for roll only, the rest are just fixed control surfaces (and back-up in case I need to get more control in mid-flight).

The 2 back tail fins are only for pitching, and the rudder is yaw only.

This way I get some stable control over the plane, and in mid-flight if I feel pitching is slow or roll is slow, I can assign more control surfaces from the main wings to give more control for the lacking part.

Yeah, I realise that doesn't help, but turns out my issue was much worse than that - had my control surfaces (on the same wing!) moving in opposite directions and working against each other. Thought it was something screwy about detecting position relative to centre of mass when attached to a swept wing, but that would only explain the pitch getting inverted, not the roll, I think? Seems massively inconsistent as well - can't figure out what;s causing some of them to be inverted some of the time. Even managed to build one where trying to roll just made a pair of ailerons move in the same direction - i.e. making it just pitch and not roll at all.

And no real idea what's causing it or how to get around it except maybe don't build such small swept delta wings...

[baldamundo]

For ....'s sake. Finally manage to make this plane have all of the control surfaces working correctly, then as soon as the fuel tanks get half empty the wings' pitch and roll controls immediately flip into inverse and it spins wildly out of control.