Spacecraft tumbles out of control

[Starstrider42]

I've found a good demo for RemoteTech's attitude control problems. Tested with KSP 0.23.5, running RemoteTech 1.3.3 + Community Hotfix, Deadly Reentry 4.7, Kerbal Engineer Redux 0.6.2.4, Module Manager 2.1.4, and Toolbar 1.7.1.

To reproduce:

• Open the persistence file at https://gist.github.com/Starstrider42/7aeaa85737fa9be41a94. Kerbal Engineer Redux should be the only mod that introduces a part incompatibility, I'm only using stock parts aside from that.
• Go to the tracking station and load the ship "Cluster OmniSat".
• The persistence file will already have a maneuver node saved. Open the flight computer and ask the ship to align with the node.

Although the ship starts out close to the direction of the maneuver node, the flight computer will start moving in a circle around it. The circle will get larger and larger until the ship is tumbling. It certainly won't ever stabilize on the node like it's supposed to.

Unlike in other instances of RemoteTech attitude control problems, you can't cheat the flight computer into submission by alternating between 5× and 1× warp -- it appears that the actual rotation is inconsistently calculated.

[wms]

@Starstrider42 Do you know if the vessel exhibits the same problem without the Engineer part attached? I'm certain it will, but wanted to be sure...

[Starstrider42]

I'll look into it.

Just realized this may be a duplicate of #25. If so, then the root cause will be multiple probe cores on the same ship.

[Starstrider42]

Ok, it's confirmed. Removing the KER module does not change the behavior. Removing all but one probe core does.

New steps to reproduce:

• Create a rocket with multiple probe cores. I've only seen this happen with Stayputniks so far, but it's likely broader. Speculation: the probe cores need to be off-axis; I don't recall seeing this bug with this satellite stack. (See next comment for a specific craft file to use)
• Launch the rocket. Let it go straight up.
• Have the flight computer point to surface prograde (hitting SRF should suffice).

Correct behavior will have the rocket roll around its long axis until it reaches its preferred roll angle. This bug manifests as the rocket deviating from vertical.

I'll try to look into this more this evening (PDT).

[Starstrider42]

Minimal working .craft example. And I do mean minimal, remove any of the four parts and you don't have a rocket anymore.

The key was making the fuel tank the root, and rotating the probe core by 90° (by pressing Q) before attaching it. As-is, the rocket can't be controlled by the flight computer. If you reattach the probe core in the default orientation (by pressing E), it works fine.

So it actually has nothing to do with multiple probe cores, those are just when you're most likely to have a non-command root, along with funny orientations for your cores.

[Starstrider42]

Looks like I was wrong. This bug is similar to, but separate from, #25.

[erendrake]

fixed with #114 ?