Aerodynamically stable EDL

[sevenperforce]

1 minute ago, Spricigo said:

The craft started to flip before 30km, that seems to be the point where aerodynamic forces and reaction wheels gave up. Your SAS is desperately trying to point you back to the desired orientation without success resulting in the spin.

The balance of weight and drag is still off. You can try to move fuel around to adjust the CoM, but better yet is to try to make it remain unaltered regardless of fuel level and loaded cargo.

Get rid of the vector, use 4 darts. This will put your CoM right in the middle of your vessel (provided you keep the cargo bay also there) which in turn will make balancing the drag much more easily.

Actually, that was happening at about 4 km. Anyway, starting with the canards feathered up for entry and then down for descent solved the nose-down problem completely, if you look at the most recent photo examples.

I've got entry and descent licked; the only remaining problem is the actual landing approach. I go from level, stable flight with no control input other than SAS:

...to opening the bay doors, still stable despite a little added drag:

...to throttling up the Thuds slowly, with corresponding instability:

...and then suddenly I roll to one side (which makes no sense; the gimbaled Thuds should have tons of roll authority):

...and dropping like a brick:

Maybe the Thuds are trying to control yaw and are throwing everything out of whack. I'll try it again with yaw authority off on all engines and roll authority only on one pair of the tail engines. I'll need to account for the fact that roll is yaw and yaw is roll for this control scheme, incidentally.

[AeroGav]

40 minutes ago, sevenperforce said:

Hmm, I had been trying to punch the vertical engines, suicide-burn style. I'll try throttling them up more gradually and seeing if I can do a more controlled landing approach.

Yeah, if the nose angle is locked by SAS,  you will see that as speed decreases, so does lift, and rate of descent increases.  With a constant nose angle this leads to rising AoA.    If you throttle up so as to keep the velocity vector constant with respect to the nose angle,  rate of descent will be under controlled and aerodynamic forces should stay within limits.

 

[sevenperforce]

6 minutes ago, AeroGav said:

Yeah, if the nose angle is locked by SAS,  you will see that as speed decreases, so does lift, and rate of descent increases.  With a constant nose angle this leads to rising AoA.    If you throttle up so as to keep the velocity vector constant with respect to the nose angle,  rate of descent will be under controlled and aerodynamic forces should stay within limits.

With yaw authority turned off for the Thuds and roll authority turned off for all but the rearmost pair, I can now throttle up gradually without flipping over. I think that attempts to correct minor variations in yaw were misaligning the Thuds and causing it to flip.

However, on the landing approach I am telling SAS to hold the internal docking port to retrograde, so that it will try to make the belly face prograde. So constant nose angle isn't the problem.

Since nose-down is my main problem right now, I think I'll try turning off pitch authority for the rearmost Thuds and see if that helps.

EDIT: Going to try some Grasshopper tests:

 

Edited May 18, 2017 by sevenperforce

[bewing]

Telling SAS to point backwards generally doesn't work very well. If you think it though, any engine that is on the wrong side of the CoM when you do that will try to move in the opposite direction from what you want. So it's wildly unstable.

 

[sevenperforce]

13 hours ago, bewing said:

Telling SAS to point backwards generally doesn't work very well. If you think it though, any engine that is on the wrong side of the CoM when you do that will try to move in the opposite direction from what you want. So it's wildly unstable.

The engines were on the correct side of the CoM, but there was a bigger problem -- the engine plumes were impinging (just barely) on the cargo bay doors. It wasn't enough to overheat them, but it was cutting my effective thrust to almost zero.

It's a shame that KSP doesn't allow you to use differential thrust in place of gimbal when you have a large engine cluster or a situation like this. I realize that with the current engine placement, there's virtually no meaningful pitching torque on the ship as a whole. Differential thrust would be much more effective because the horizontal distance from CoM would be an advantage instead of a hindrance.

I'll try rearranging the engines, cargo bay, and fuel tanks so that CoM is 100% balanced at all times during the landing burn.