Once this plane is rolling, it won't stop.

[Laie]

For a challenge, this plane is supposed to fly fast (very fast) at a high AoA (example). Hence the vertical Mk2 parts. Unsurprisingly, these cause trouble in the yaw axis, which necessitates the huge tail fin. And now, that is creating problems of it's own.

Trouble starts when I want to slow down for descent. To do that, I pull up for braking, while trying to control the descent by rolling. Here's a demo of that maneuver during a more leasurly spaceplane entry, but of course, for the sake of the challenge it happens a wee bit faster and more aggressively.

I'm not sure if my problem is proper inertia coupling, but at any event, rolling the pitched plane induces some yaw, which grabs me by the tailfin and adds more roll force in the direction I'm rolling anyway. Positive feedback... and once it happens, I can't snap out of it:

Just look at the size of those pointers. I don't think I can reasonably add as many control surfaces as would be necessary to counter that.

I consider adding a rudder close to the centerline, or possibly even below. Which is a cheesy kerbalism, but if that is what it takes...

Still, I'd like to ask: beyond MOAR CONTROL, is there anything I can do in order to make it inherently more stable? Or at any rate, less unstable?

[OHara]

Lowering the tail-fin is (as you said) the direct solution and probably the most effective.  I have often put two small vertical wing-squares on either side of the engines on a spaceplane.  The more commonly-seen real solution is to split the tail-fin into two less-tall tail-fins.

If you want to look at traditional aeronautics solutions, people call your symptom 'spiral instability'.  The classic solution is a dihederal angle on the wings, but you probably don't want that here. 

[Signo]

In my experience the answer Is "no". I had the same (or very similar) issue with a mk3 craft with a couple of side mk2 vertical "gondolas" and the only way I found to minimize the roll was to add a tail fin mig-23/27 style as an extra yaw control. This however gave me landing issues (the lower fin was doomed to crash at any landing). I posted a question around here about the issue a few years ago and the only interesting consideration I can remember was about "torsion of wings due to overpower", so maybe "moar struts/autostruts" could help. Good luck, I read the challenge topic, It seems fun.

Edited January 1, 2020 by Signo

[bewing]

How tall are your landing gear at the front? You'll get moar control by adding an additional set of canards at the front than you will by doing anything else with control surfaces. With your current design, I'd mount them vertically and displace them toward the centerline a bit. But that probably won't work with your landing gear. So the next choice is to place two sets of canards in a reasonably shallow X on the nose (so that they don't touch the ground during liftoff -- they need a kerbal cubit of clearance or they will scrape the ground). That will get you an overabundance of pitch control at the front, but it will also get you some yaw control, which sounds like the thing you need most. And then you can play games with turning that cwazy tailfin into perhaps 3 or more smaller tailfins at the back end.

Edited January 1, 2020 by bewing

[Laie]

12 hours ago, OHara said:

If you want to look at traditional aeronautics solutions, people call your symptom 'spiral instability'.  The classic solution is a dihederal angle on the wings, but you probably don't want that here. 

Thanks for the hint, that got me thinking...

I guess I'm doing it wrong: I control the plane solely by pitch and roll (well I don't, but the kRPC PID controller does). There is zero yaw control on the vessel because the PID controller doesn't know about pitch/roll steering, and, given any control, will only make hash of it.

That means that I set p/r and leave it to aerodynamic forces for yaw to follow suit. This works well until I crank up AoA for braking, when suddenly it goes all wrong. A simple look at the navball tells me that rolling at AoA always requires some yaw moment, and well, I guess I will have to provide it.

15 hours ago, Signo said:

add a tail fin mig-23/27 style [...] was doomed to crash at any landing

13 hours ago, bewing said:

How tall are your landing gear at the front?

I guess I could crank it up using the next size landing gear... However, it just occurred to me that a yawing canard at the front/top will induce roll in the desired direction. Let's see how that works out:

Turns out that it looks just too silly to have it hang out at full length, so I sank it into the fuselage anyways. But: it works!

The plane still spazzed out a few times, but with the new canard it quickly recovered. "Some tweaking required" ... compared to the trouble I had before, this is a resounding success.

Thank you everybody!

[UncleManuel]

I just realized now that these are Mk2 parts - just turned 90° on their sides... 

With this hilariously overpowered design you should watch the CoM, CoL and CoT indicators. And for good measure at full and empty tanks. Even on reentry planes are sensitive to CoM shifts... 

You can also try to replace the HUGE fin with a standard canard, deactivate its roll and pitch control and turn on S.A.S. for automatic yaw control. Because the gimbal of the engines doesn't have much authority up there...

[OHara]

17 hours ago, Laie said:

I guess I'm doing it wrong: I control the plane solely by pitch and roll

Pitch and roll certainly sounds like the correct way to control a  plane.  (Maybe my irony-sensor is failing me again.)

I know you have a moar-control solution for this craft, but I think you will like trying passive solutions in the future. 

Since the only valid reason to use rudder pedals, is to counter adverse yaw induced by using ailerons to roll (or to be draggy on purpose)
you could program your KRPC roll control to also give the coordinated rudder input (opposite direction to what enabling the default roll-response on the rudder would give).

I like making upward-only deflecting ailerons.  If you use the SPH's rotation tool to deflect the ailerons to full up, and then 'deploy' them back to neutral, leaving them always 'deployed', then when rolling in flight the aileron on just one wing will deflect up to drop that wing, giving just a little yaw in the favorable direction.

Less induced adverse yaw, smaller vertical stabilizer (tail fin), less spiral instability.

Edited January 3, 2020 by OHara
and I am going to reference your clever vertical mk2's in my feedback item on mk2 drag https://bugs.kerbalspaceprogram.com/issues/23989

[Laie]

9 hours ago, OHara said:

Pitch and roll certainly sounds like the correct way to control a  plane.  (Maybe my irony-sensor is failing me again.)

Well yes, but I've gone so far as to not provide any yaw control, relying on aerodynamics to pull my vessel around. Which works fine most of the time, mind you... but not at the high AoA I want for entry.

Just to throw out a few numbers, it's fine on it's own up to 10° AoA. With the canard, I can crank it up to 30° -- which is less than I'd like, but all it can do even with fuel trimmed to the tail.

9 hours ago, OHara said:

coordinated rudder input

I considered that, briefly, but the rub lies with the word "coordinated". Especially as the vessel goes from one ridiculously extreme situation to another in a matter of moments. In the end it's easier to watch and control the pitch/roll/yaw axes individually.

 

[Snark]

On 1/1/2020 at 10:08 AM, Laie said:

I'm not sure if my problem is proper inertia coupling, but at any event, rolling the pitched plane induces some yaw

Did you disable roll authority on the tail fin?  Because if you didn't, the tail fin will try to "help" you roll, which is exactly precisely what you don't want.  With roll authority turned off, you should never get any yaw result when rolling.  (The reverse-- i.e. getting a bit of unwanted roll when you're trying to yaw-- is basically unavoidable with a lone tail fin, but that's generally not a problem assuming you've got some well-placed ailerons on the wings to compensate-- they're a lot farther from the roll axis than the tail fin is, and therefore have a lot more roll authority.)

(Apologies if it's a silly question, but a lot of folks forget to do that.) 

[RizzoTheRat]

Assuming kPRC works anything like kOS, can't you have a separate PID (or possible just P) controller that just looks at the yaw ( ie exclude the ships top vector direction) angleand tries to keep yaw angle to 0 independently of other directions? 

 

As for roll stability,  some anhedral on the wings would help, but might cause you com/col issues at high angles of attack. 

[Laie]

On 1/5/2020 at 9:31 PM, Snark said:

Did you disable roll authority on the tail fin?

If you look closely, you notice there is no tail rudder. Originally this vessel had now yaw controls whatoever, cf the reply just above yours. The problem is solved, btw. I'm hard put to rate any answer as "best", but the first day's replies gave me enough food for thoughtm abd eventually I put on a single canard at the front for yaw control. Which is shaky, but possible.

3 hours ago, RizzoTheRat said:

can't you have a separate PID (or possible just P) controller that just looks at the yaw

Just like that. I'm taking my cues from what kRPC reports as sideslip and deploy the control surface by minute amounts, bypassing ordinary pitch/yaw/roll input.

I've since started doing the same with the elevons, for coordinated rudder input as @OHara suggested. I guess I should go all the way and look into pitch as well.

[Snark]

2 hours ago, Laie said:

If you look closely, you notice there is no tail rudder. Originally this vessel had now yaw controls whatoever, cf the reply just above yours. The problem is solved, btw. I'm hard put to rate any answer as "best", but the first day's replies gave me enough food for thoughtm abd eventually I put on a single canard at the front for yaw control. Which is shaky, but possible.

I also note that your ailerons are really close in to the central roll axis of the plane, due to the stubby wings.  That means it's going to have very little roll authority.

If it's solved, great, but in general:  if you have an issue with insufficient roll authority, it's good to move the ailerons (i.e. roll control surfaces) as far out from the axis as possible to give them a better lever arm to work with.