Stability Help

[Carl]

So i posted my tourist 10 man design over in the exchange, (Link), And upto now using Mechjebs ASAS i've had no issues getting her to orbit. I've just unlocked the accent guidance however and it's revealed somthing, the ship is incredibly unstable in roll at low altitudes. This has never been an issue before as my old launch method, and probably won't be an issue for most people using it. However now i want to use mechjeb's accent autopilot i need to sort this out. The problem is i have no idea why it's unstable. with 16 control fins it should be rock solid stable, (as it's behaved in yaw and pitch). if anything a sever lack of reaction wheels should be causing any problems that do occur. So does anyone have a clue why it's being so unstable?

[Magzimum]

We're talking about this ship (there's no shame in posting it again):

I will gamble a few first impressions:

• Drag should be as far to the bottom as possible. So the fins/wings are too high up. The 4 fins at the very top are probably your worst offenders. They can only be used to turn the ship away from prograde very efficiently while going up (which you don't ever want), or for stability during descent (which is not the problem). So they serve no purpose during ascent and make stability much worse. Such canards are common on stuntplanes or jet fighters, but those are designed to be unstable! Rockets pretty much never have canards.
• The other fins at the top of the boosters are also much higher than on conventional designs. 
• You appear to have placed the parachutes only on one side of the passenger modules. They actually have significant mass (0.1 ton each, I believe) so your center of mass is not in the center anymore. That means that your rocket will continuously pull to one side, which must continuously be compensated by the wings and rocket engines (gimbals).

 

[Carl]

Well first if i don't have the wings on top it ain't getting down from orbit. It will either burn up or arrow into the ground too fast for chutes without them. It's natural descent orientation is to point the nose of the command pod along the air flow direction and just fly nose first, which isn't up to taking the heat and really doesn't bleed enough speed.

 

Likewise the ones on the liquid tanks are there because without them, no control past 2km altitude when the SRB's pop off. the gimbals on the Liquid engines are nowhere near enough, in fact i had to disable roll on them because they made it worse.

 

The Chutes. I can try fiddling with those. The issue is it's no good if it comes down on it's tail or it's nose, an angle is fine but if it tries to balance on the tail it's a problem because the chutes disappear, it tips, and boom.

[Carl]

Hmm good news, your right on the fins, stripping the lot and she seems to go up well, though control authority is poor past about 30km. However without the top end fins i'm having the same getting down sisues. I guess the need for the lower fins came from the drag from the upper.

[bewing]

Well, I disagree with Magzimum. I put canards exactly like that on the tops of almost all my rockets for several reasons. Yes, positive aerodynamic attitude control is naturally unstable but the benefits outweigh the drawbacks. Especially on reentry, as Carl says, when your RV becomes more of a spaceplane glider instead of an overpowered rocket.

As far as the roll stuff goes, I'd think the problem & solution is pretty simple. It's really easy to make rockets (or planes) roll, compared to making them turn in the other 2 directions. Sadly, you can't adjust the "control authority" in just one direction at a time. I'll assume that you have to have about that many control surfaces in about those places to make the thing fly in pitch and yaw. But what you can do is completely disable roll authority on some of them. I would bet that the only offenders are actually the tailfins on the liquid tanks. Disable roll control on those, and I bet you'll be good to go.

In more detail:

To figure the roll torque from a control surface, you look at the lever arm to the centerline of your ship. The canards at the front have a short lever arm, so they make small roll torques -- which is good. The tailfins on the liquid tanks have much longer lever arms in that dimension, so they make lots of torque. So those are the bad ones.

The funny thing is that in the other two dimensions, those tailfins on the liquid tanks should be very near the CoM, so they should have almost no control authority. So I'm really surprised to hear they do anything good. So besides turning off their roll authority, I'd certainly try cutting their number to 3. I'd try moving them way down to the bottom. I'd try replacing them with the AV-R8 surfaces instead.

I understand that the liquid tanks are placed with 6x symmetry, so the editor naturally wants to use 6x symmetry for anything you attach to them. But if you place the winglets one at a time, you can override that. Yes, I understand that if you place them one at a time, they might not be perfectly symmetrical. But you can use the markings on the tanks to get them lined up very close to symmetrical.

 

 

[Carl]

Quote

The funny thing is that in the other two dimensions, those tailfins on the liquid tanks should be very near the CoM, so they should have almost no control authority. So I'm really surprised to hear they do anything good. So besides turning off their roll authority, I'd certainly try cutting their number to 3. I'd try moving them way down to the bottom. I'd try replacing them with the AV-R8 surfaces instead.

Well the thing to bear in mind is under the old take it up manually with MechJeb ASAS was that the ship tended to wobble with just the canards at the top. But with the extra fins thee ASAS was doing very little in the way of control inputs that i could see, so i suspect they were working like fixed fins and dampening yaw and pitch by being there. They also helped during the initial turn to get the ship pointed where it needed to go, i suspect the gimbaling on the engines was interacting with them and they became a fulcrum for the gimbals, i just know stripping them all has made it handle worse in the 30-40km range.

 

But yes, without the fins on the upper stage i'm having a lot of issues getting it down safely again. Gonna try putting another heat shield on the top of the command module, see if that helps. Should fix the heating issue i think, but getting the velocity down is gonna be a pain.

[GoSlash27]

1 hour ago, Carl said:

Hmm good news, your right on the fins, stripping the lot and she seems to go up well, though control authority is poor past about 30km. However without the top end fins i'm having the same getting down sisues. I guess the need for the lower fins came from the drag from the upper.

I'd just like to point out that fins are worthless above 27km altitude. If you need more control authority up there I'd recommend reaction wheels or RCS.

And yeah... I imagine you're going to have a very hard time getting that long spindly shape to behave during reentry.

 

I think your original roll instability was from arranging your tailfins in a multiple of 3. If you put fins in planes that aren't aligned with pitch and yaw, they will induce cross- coupling every time they move.

Good luck,
-Slashy

[monstah]

3 minutes ago, GoSlash27 said:

I think your original roll instability was from arranging your tailfins in a multiple of 3. If you put fins in planes that aren't aligned with pitch and yaw, they will induce cross- coupling every time they move.

Really? That makes sense, but I've been using 3 fins on almost every rocket since I got this game and it has never been much of an issue. I guess it's only a problem if you already have other design concerns?

[GoSlash27]

Monstah,

 I'd suggest trying it with 4 so you can judge the difference. I'm occasionally forced to use 3 fins in early career due to part count, and I notice the weird behavior immediately.

Best,
-Slashy

[Carl]

Quote

I'd just like to point out that fins are worthless above 27km altitude. If you need more control authority up there I'd recommend reaction wheels or RCS.

It could be the high fin to mass ratio i had but before the fins seemed to work upto about 35km.

 

Think i've got it licked though now, putting the heat shield on the command pod nose and bringing her down nose on seems to have fixed most of the issues, it even naturally swings the cheeks out at around the right altitude/speed to aid speed bleed off. And yeha i put some reaction wheels on as i was getting in orbit authority issues too. Tight on weight and still having issues getting the chutes to set her down side on.

[Spricigo]

Some ideas to deal with the tumble&explode problem:

-set up some bumpers (e. g.  girder segments around the cabins) 

-put parachutes in both sides of the rocket but deploy only the chutes in one side (or cut some after they did the job and before you land) 

-use a tiny engine with just enough power and propellant so you can use it to help you turn to horizontal position just before landing. 

 

Keep in mind those are rather crude solutions,  so use it only if you really need your craft working NOW and don't stop to look for better engineering. 

[Carl]

Ehh, main problem is i have less than a ton of weight to work with, got the panels sorted, now she's coming down nose first again though on the chutes. ughhh. I will get there.

[Carl]

Nailed It!

 

Check the original linked thread for the craft file, and here's a copy of the pics. Notre i added a couple of chutes and a pair of batteries to the final stage after this flight, but they're detail changes only, and are more just in case. Also uploading all those images was a massive pain, imgur was being really buggy with me, would constantly get stuck "processing" :(. Guessing the servers are overloaded or somthing.

 

Pictures:

 

Yes, as you can see below at zero inclination and with mechjeb you can get it to orbit with 143m/s left in the liquid boosters. That's a bit of a special case though, at any other inclination or manually you'll likely get there with just a few hundred left in the final stage.

 

 

[klesh]

Nice work dude.  This is what KSP is all about.  Building what feels groovy, identifying potential design flaws, and working them out one by one until you can achieve you self-appointed goal.

 

That delivers some of the most satisfying gaming I have been able to find.

[Carl]

Heh, thanks, i'm going through version 2.0 of this with my Mun Bus atm. Having some issues keeping the upper stage d/v where i need it whilst getting a stable design that will lift it.