Roll issues during high-altitude staging

[MaverickSawyer]

During my second rocket launch in 0.17 (first one was a simple once-around test), I was aiming for a Munar landing. Launch goes fine until I jettison my first set of liquid boosters at about 15 km. I had paid very close attention to their alignment on the decouplers during assembly, because I know that having them off even slightly can impart some serious roll issues to your craft. I also had added winglets to maintain control. No dice. Fuel drained, staged, and the craft began to accelerate it's roll. any thoughts?

[TG_bigboss]

sas would help. I know i have had this issue. It has to do with the center of mass and thrust.

[TheDarkStar]

Winglets don't help much at those altitudes. The best thing to do is to use struts, lots of them.

[phoenix_ca]

Struts will help with stability, but they won't cancel any angular momentum that's imparted to the rocket.

Could you perhaps post a picture of the rocket? I've had some serious roll issues myself. My successful designs had...fewer parts? Maybe that's it? Ish? I'm not sure about that at all.

[vexx32]

The roll issues can all be solved by attaching multiple standard SAS modules to the rocket. The more you have, the more control authority you've got.

[karolus10]

I think that any rocket design that cannot be handled without SAS and auto-pilot, need go back to drawing board .

[Xeo]

I think that any rocket design that cannot be handled without SAS and auto-pilot, need go back to drawing board .

Try to fly a rocket without ASAS with your keyboard ... and change your mind ^^ Plus perfectly symmetrical designs sometimes start rolling without any reason, and SAS units are more than useful in these cases.

[karolus10]

I fly mostly rockets on precision mode and without any ASAS - using my beloved IBM keyboard, so I don't see any problem .

[Khrissetti]

I've had the problem since .13. I find just toggling the ASAS off and on again as soon as you seperate cures it.

[phoenix_ca]

Sure, except I've had symmetrical rockets that had oodles of SAS modules strapped all over the place (from NovaPunch, 3 per stage, plus two per booster), and they still rolled out of control. I had seemingly no control authority for roll, but incredible ability to adjust pitch and yaw. It felt like some external force was acting on the rocket to make it roll.

[Zephram Kerman]

ASAS tries to return the craft to whatever orientation it had when it was turned on. This includes the roll angle. So if there is very little control authority in the roll axis, it will overshoot several times and take forever to actually stop. If you toggle it off and on again, this resets the target orientations. (You can also tap "F" to do the same thing.) Do this several times for a simple damping effect.

Sure, except I've had symmetrical rockets that had oodles of SAS modules strapped all over the place (from NovaPunch, 3 per stage, plus two per booster), and they still rolled out of control. I had seemingly no control authority for roll, but incredible ability to adjust pitch and yaw. It felt like some external force was acting on the rocket to make it roll.

I assume you're using the symmetry tool? That helps a lot. Otherwise, it sounds like another effect (which I don't know the name of). It's basically unintentional roll forces caused by pitch and yaw controls having a side-effect. A good example of this problem is the Albatross that comes with 0.17; it's hard to turn, because the rudders activate when you're trying to use ailerons, and vice-versa. Try it with ASAS turned off, and see if it happens any differently.

Another possibility is physics simulation weirdness. (Yes, that's a technical term.) When you load the vessel in VAB, disconnect the first part from the command pod. Put it down. Pick it up again. Reconnect to the pod. Doing this reestablishes all the connections between parts, and eliminated some weird problems in 0.16 and earlier. I haven't used 0.17 enough to know if this still matters, but it's an easy thing to try.

Edited September 24, 2012 by Zephram Kerman

[MaverickSawyer]

Tried all those suggestions BEFORE I came here. Deactivating the ASAS only caused the roll to accelerate, as did reactivating the ASAS. I was also using the LVT-45 engines, which are gimballing, so I should have had the requisite roll control. I'll post pics and a ship file later.

[Teh_Miner101]

Use some SAS modules (not ASAS) throughout your ship. They should help stop the rolling.

[MaverickSawyer]

OK, rocket looks like this:

First/outer stage is six boosters comprised of 4 400 liter tanks and a single LV-T30 engine per booster. These fire at the same time as the inner ring of six boosters, comprised of another 4 400 liter tanks and a LV-T45 engine each. The outer booster cross-feeds to the inner booster on each "arm" of the vehicle. The core is 3 400 tanks feeding to a single NTR. This is lit only at altitude.

Issues occured during flight. I will detail the flight path:

During the ascent, I remained straight until about 10 km. At that point, I tipped about 15 degrees toward the horizon and locked the ASAS. No issues yet. At about 15 km, and 173 m/s, the outer boosters run out of fuel, and I cut them loose. At this point, the roll begins, rapidly accelerating. This is with the fins locked at full deflection in the direction to counter the roll, and RCS active on the lander stage above. A total of 18 fins are ineffective, as is the full gimbal of the engines. I kill ASAS, and the roll accelerates more. Reactivating the ASAS make no difference. I shut down all thrust, stage repeatedly until I am on the Kerbin Return stage, and burn just long enought to clear the mess of spent stages beneath me, and then shut dow the engines, seperate the capsule, and deploy my parachute. End of flight is in the ocean off the KSC.

The attached ship is slightly modified from the shown rocket, as it has some more struts, and I removed the fins from the NTR stage. Feel free to take it for a spin, and confirm or deny my results.

Ship: [ATTACH]33485[/ATTACH]

[pebble_garden]

I've noticed this as well.

I haven't done *extensive* testing, but my sense is that rocket design techniques I used with great success in v0.15-0.16 now experience more roll issues, in the flight phase described by MaverickSawyer.

I don't know if it's due to changes in the physics code or something else, or if I'm just imagining it. But I take great pains to keep everything symmetrical, and I still have more trouble with roll. It's usually manageable, but it makes me wonder what I'm suddenly doing wrong.

I'd prefer not to mask design flaws with more attitude control, if I can help it. So I'm trying to find better ways of assuring perfectly symmetric strut placement, like putting down reference strut heads on the snap-to angles to help guide the placement of the strut tails (which don't respect snap angles).

That said, I'm hardly complaining. I am very happy with the tweaks that have been made to the ISP of the various fuel tanks, as it's much easier to get in orbit that with the unofficial fuel consumption bug hotfix for 0.16.

Edited September 24, 2012 by pebble_garden

[MaverickSawyer]

OK, redesigning the rocket using MechJeb instead of ASAS to see what it thinks of the design. MechJeb+real Jeb=Success? We'll find out here soon enough...

[Werwolf]

Make sure you have plenty of regular SAS, and you only really need one advanced SAS. You can have five hundred SAS modules, without at least 1 ASAS to "master" them, all the SAS modules in the world are useless. Only having ASAS is also useless, except for atmospheric control via canards.That's been my experience in testing, I could be horribly utterly wrong though.

[Ghery88]

My crafts roll too, but most of the time I can handle it fairly early. Gimbaling engines, well placed RCS and ASAS are used.

[vexx32]

Looking at that craft... try slapping on a standard SAS module on top of each of those lower stages, where there's room, and one each on top of the RCS tanks as well.

[MaverickSawyer]

If I had space to do so, I would. However, the tanks are just clear of the lander engines, so no dice. as for the RCS tanks on the lander, I fail to see any point of adding them there...

Wait a sec...

Would those tanks cause roll issues if they are swaying back and forth around the long axis of the rocket? I remember watching those tanks rocking and rolling during roll oscillations during the first part of the ascent. That's why I added an additional 4 struts to the posted craft.

[phoenix_ca]

ASAS tries to return the craft to whatever orientation it had when it was turned on. This includes the roll angle. So if there is very little control authority in the roll axis, it will overshoot several times and take forever to actually stop. If you toggle it off and on again, this resets the target orientations. (You can also tap "F" to do the same thing.) Do this several times for a simple damping effect.

I assume you're using the symmetry tool? That helps a lot. Otherwise, it sounds like another effect (which I don't know the name of). It's basically unintentional roll forces caused by pitch and yaw controls having a side-effect. A good example of this problem is the Albatross that comes with 0.17; it's hard to turn, because the rudders activate when you're trying to use ailerons, and vice-versa. Try it with ASAS turned off, and see if it happens any differently.

Another possibility is physics simulation weirdness. (Yes, that's a technical term.) When you load the vessel in VAB, disconnect the first part from the command pod. Put it down. Pick it up again. Reconnect to the pod. Doing this reestablishes all the connections between parts, and eliminated some weird problems in 0.16 and earlier. I haven't used 0.17 enough to know if this still matters, but it's an easy thing to try.

That's...a neat trick. I might give it a try. I'm indeed using symmetry, but I hadn't considered the cumulative effect that pitch-and-yaw oscillations that are caused by both MechJeb and ASAS would have on roll. Any rocket that I've had roll problems with also suffered from a lot of oscillation-induced-wobble around the center of mass.

[Zephram Kerman]

...Would those tanks cause roll issues if they are swaying back and forth around the long axis of the rocket? I remember watching those tanks rocking and rolling during roll oscillations during the first part of the ascent. That's why I added an additional 4 struts to the posted craft.

If I understand correctly, yes. Try putting some struts diagonally, and from the lander to somewhere on the boost stages.

[MaverickSawyer]

Well, I've decided to scrap that design and start fresh. BTW, MechJeb really locked down the oscillations. AKA not longer visible.

[Ghery88]

Watch it with MechJeb, the ascent autopilot broke my craft apart before getting to the rolling issue.

[MaverickSawyer]

I know. I've had designs that started the gravity turn, and just kept going... X_X