Spaceplane build tips needed - plane is never "roll stable"

[Plusck]

Hi, I'm much more experienced in rockets than planes, but I wanted something to enter Laythe, look for landing place, land, and reach orbit.

This thing just does that. It's really very very simple.

Spoiler

Craft file is here: SP01 simple 01deg v2.craft
It has wing AoA of 1 pixel rotation (is that a degree?), control surfaces and landing gear rotated back to 0°, very little clipping other than the pre-cooler which is offset into the adapter tank.
It flies nicely. About 15 "presses" of elevator trim for takeoff and for the climb up to 9km. Down to about 8 presses of trim to hold a prograde angle of about 8 degrees on the start of the speed run, reaching 1450 m/s at 24km ready to switch to closed cycle and - then an only then - activating SAS. And for re-entry, maximum vertical trim keeps it at a decent angle. It makes LKO with very little spare LFOx remaining and about 100-160 LF spare for atmospheric flight.

Spoiler

pic in orbit

However, I can never, never get it to be "roll stable". I dislike SAS intensely, so I try to fly only with trim and fine control. And that works fine apart from the need to correct the roll every 4 seconds.

Is there any way of getting a plane to fly straight without SAS?

Are there any requirements for build order (i.e., place each wing part and control surface individually with symmetry, or build the whole wing and then place it as one piece with symmetry? I seem to remember there being a build order required for multiple engines and air intakes...)? Does the landing gear have an effect on stability (I'm pretty sure it is perfectly aligned, at least is gives no other problems on takeoff or landing)?

I've tried using a rocket fin rather than the tail fin shown, but that didn't make any difference. I'm pretty certain the cargo is balanced.

Help! Please!?!

[OHara]

Persistent roll, always in the same direction, seems to be cause by wings flexing asymmetrically, even if the design is nominally symmetric.  Placing wings directly off the root central part has helped this for me (you seem to have done that already) and struts helped when I had heavy engines on the sides.

After the persistent roll is reduced, though, there is still some residual error that makes the plane wander. 

Dihedral angle in the wings should bring the plane back to level.  A v-tail converts roll into a yaw error, while leveling the plane, but worked well for me.

Edited September 5, 2016 by OHara

[fourfa]

I've seen asymmetric aero lift/drag/control/etc problem arising from placing wings on cargo bays, then offsetting for cosmetics.  Somehow the game sometimes find the wing to be "shielded" by the bay on one side but not the other.  A sure-fire diagnosis for this is if you crack open the cargo bay for a second, the problem goes away.  Then it usually keeps symmetry for a while.

[Plusck]

5 hours ago, OHara said:

Persistent roll, always in the same direction, seems to be cause by wings flexing asymmetrically, even if the design is nominally symmetric.  Placing wings directly off the root part has helped this for me (you seem to have done that already) and struts helped when I had heavy engines on the sides.

After the persistent roll is reduced, though, there is still some residual error that makes the plane wander. 

Dihedral angle in the wings should bring the plane back to level.  A v-tail converts roll into a yaw error, while leveling the plane, but worked well for me.

Well, I tried that craft but it was difficult.

Maybe my install has something wrong with it, but the roll on that ship was extremely hard to control, and highly variable with speed. After a couple of tries I made orbit... at 70.2 x 70.1 km and with 12m/s left in the tanks.

Still, I did re-root the craft...

3 hours ago, fourfa said:

I've seen asymmetric aero lift/drag/control/etc problem arising from placing wings on cargo bays, then offsetting for cosmetics.  Somehow the game sometimes find the wing to be "shielded" by the bay on one side but not the other.  A sure-fire diagnosis for this is if you crack open the cargo bay for a second, the problem goes away.  Then it usually keeps symmetry for a while.

Aha. The wings were indeed attached to the cargo bay.

So, I swapped the two rear tanks out for the long adapter tank. Attached the wings to that, rotated to align and offset them fowards. And rerooted to that tank. And offset the wings upwards by three pixels.

And it worked!

I wasn't being scientific about it so I don't know which definitively helped, but now I only have to correct the roll every 10-15 seconds rather than every 3-4.

And I managed to get to orbit with 300m/s LFO remaining + 160 units LF.

Spoiler

Improved craft file

Yay!

Now it's going to Laythe.

[OHara]

3 hours ago, Plusck said:

Well, I tried that craft but it was difficult.

Maybe my install has something wrong with it, but the roll on that ship was extremely hard to control, and highly variable with speed. After a couple of tries I made orbit... at 70.2 x 70.1 km and with 12m/s left in the tanks.

That was a wobbly early-career SSTO using Whiplash and LV-Ns, in which I had the rolling problem you described and did the best I could to resolve it with aerodynamics.  When the roll did wander off, I tried to make the plane 'Dutch roll' rather than spiral, so the roll didn't build up too quickly.

That ship had no fuel lines; you probably had at least 1750 m/s of liquid fuel in the central tanks when you reached orbit!

3 hours ago, Plusck said:

And it worked!

I wasn't being scientific about it so I don't know which definitively helped, but now I only have to correct the roll every 10-15 seconds rather than every 3-4.

Your version-3 flies straight as an arrow.  I experimented a bit to try to find what it was that helped.  
I tried moving the root part (using change root' in the SPH) and that made no difference.  Mounting the wings to the tank in front of the cargo by made the problem worse.   On the version with wings mounted to the cargo bay, opening/closing the cargo bay made no difference. 

Everything I have seen is consistent with wobbling joints being the cause of the wandering roll.  Having fewer joints between the heavy engine and the wings makes the plane fly symmetrically.

[Plusck]

7 hours ago, OHara said:

That ship had no fuel lines; you probably had at least 1750 m/s of liquid fuel in the central tanks when you reached orbit!

LOL. Ok that was particularly stupid of me. I was just watching KER's readouts rather than having the resources window up.

Proud to have made orbit with 12 m/s plus 1750 then

7 hours ago, OHara said:

Your version-3 flies straight as an arrow.  I experimented a bit to try to find what it was that helped.  

I tried moving the root part (using change root' in the SPH) and that made no difference.  Mounting the wings to the tank in front of the cargo by made the problem worse.   On the version with wings mounted to the cargo bay, opening/closing the cargo bay made no difference. 

Everything I have seen is consistent with wobbling joints being the cause of the wandering roll.  Having fewer joints between the heavy engine and the wings makes the plane fly symmetrically.

I think you may be right - minimal parts between engines and wings sounds logical and fits the case. I'll have to remember that. And be far more careful with cargo bays, not something I've really had much experience with as a rocket man. Thanks again for your help on this.

 

And I also never knew that items surface-mounted to the cargo bay bulkheads are not occluded for drag, so all of my experiments were potentially causing drag (tbh, I don't know whether experiments are still dragless or not - or whether anything is in 1.1.x).

Thanks to having just seen this:

So I made a version 3b with the experiments mounted to a node-connected object in the cargo bay. Hard to know whether it was just endless attempts which have improved my flying or whether that made a significant difference, but I made 76km orbit with 325 m/s LFO + 172 units LF remaining. It certainly seemed zippier in the lower atmosphere. And it only needed correcting the roll every 30-40 seconds.

 

Has anyone made a list of all the tips and tweaks you really need to know to build planes, for KSP 1.1.x ?

[pincushionman]

Stiffer stiffer stiffer stiffer stiffer KJR stiffer stiffer stiffer stiffer struts stiffer stiffer stiffer stiffer…