Trying to cure rudder

[OlWilly 130]

I've been having problems with rudder behavior in SF2 ever since I picked up the game. It not only behaves weird, but also has very limited effectiveness. The two main problems:

- sideslips and slipturns seem to be impossible. Aircraft barely reacts to rudder movements in these aspects and just keeps flying with barely any change in heading. This means that rudder gives you only yaw angle and no slip force at all.

- the annoying "rocking", "swaying" motion. As you apply rudder (or return it to neutral) and aircraft yaws accordingly, it then starts to sway on yaw axis like a boat on the sea - before finally stabilizing. This is present in other sims too, but not to the same degree as in SF2. Moreso, in SF2 using rudder you can wiggle the aircraft on yaw like a dog's tail which is completely insane.

For digging deeper I firstly picked up MiG-15bis from download section as it seems to have the worst rudder behavior. I will post excerpts from data file with my comments.

[FlightControl]

PitchDamper=0.0
RollDamper=0.0
YawDamper=0.5
AutoTrimLimit=0.0

Initially, I set up all these values to zero for the purity of experiment. Yawdamper is at zero in majority of aircraft I checked.

Afterwards, I applied YawDamper=0.5 to damp the smaller swaying motion of the aircraft. Touching other parameters here is not required

[AircraftData]

EmptyMass=3668.0
EmptyInertia=21264.0,7800.6,1800.0

It goes as pitch, roll and yaw inertia. No need to touch the first two parameters, and the last is one of the main factors contributing to the said behavior.

I played with values a little bit.

At really high yaw inertia (I tried as much as 20 times the empty weight), aircraft sways slowly but with high amplitude, exactly like a boat.

At lower values, swaying is faster but has smaller amplitude and ends quicker - which is more like the real thing. I think values of 0.5 - 2 times the empty weight are optimal here.

[LeftAileron] / [RightAileron]

CDdc=0.0201 //drag
Cldc=0.0395 //roll
Cndc=-0.005 //yaw

ControlRate=4.0

I am not touching table data because I really have no idea about it. These coefficients are simpler to understand though.

For the clarity, positive value means the movement opposite to the direction of deflection. Negative value is in the same direction as deflection. Thus, Cldc=0.0395 on left aileron means that aircraft will roll in direction opposite to aileron deflection.

No need to touch CDdc.

Cldc has no effect on the rudder, you can play around with it if you want better or worse roll rate.

Cndc is the adverse yaw. Aircraft tends to yaw slightly in direction opposite to roll, the effect discovered long time ago by Wright brothers, so make sure this value is opposite to Cldc.

ControlRate affects the speed at which control surface moves after your input. I don't think it is necessary to change it

[Rudder1] /[Rudder2] (this model has split rudders with their effects combined, the same applies to models with one rudder)

CDdc=0.0162     //drag
Cydc=-0.12       //side force
Cldc=-0.0024     //roll
Cndc=-0.028     //yaw

CDdc=0.0123        //drag
Cydc=-0.10          //side force
Cldc=-0.0016        //roll
Cndc=-0.02        //yaw

Once again, drag is ignored

Cydc is the side force caused by slip. It should be in negative as aircraft will slip in the same direction as rudder applied. Curiously, original file had it positive and aircraft tried to slip in the opposite direction. This is the reason for lack of slip!

Cldc is the rudder roll caused by different airspeeds over inner and outer wings during slip and thus, different amounts of lift. Likewise, should be in negative as aircraft rolls in the same direction as rudder applied, and once again, in original file it was positive.

Cndc is the yawing motion and the main source of swaying. It should be negative too, otherwise, doesn't require much tuning.

With the given changes, the handling of the aircraft (at least for me) improved a lot, and gunnery became way more easier. The swaying motion remained only on minute, abrupt rudder inputs and I can finally feel the rudder working, including slipturns too.

To confirm my discoveries, I looked into the stock F-100A. It has the same behavior, persistent swaying on yawing, possible to be wiggled around violently and no slip present.

[FlightControl]
YawDamper=0.5

Applied yaw damper, rest left untouched.

[AircraftData]

EmptyMass=8226.0
EmptyInertia=68970.9,20295.1,8000.0

Original had yaw inertia well over 100000. Not sure why it is always set up so high, I set it up of around 1 time empty weight

[LeftAileron] / [RightAileron]

CDdc=0.0356   //drag
Cldc=0.0690   //roll
Cndc=-0.0075   //yaw

All values left as in original, except I turned Cndc to negative. Curiously, original had it in positive meaning that it had no adverse yaw

[Rudder]

CDdc=0.0097  //drag
Cydc=-0.1130  //side force
Cldc=-0.0008  //roll
Cndc=-0.0730  //yaw

Didn't touch the drag and yaw, but, most curiously, original had Cydc and Cldc as positive!

This means that on right rudder it tried to slip to the left which, if effect, left it flying straight.

For some reason, even with Cldc positive it rolled in the correct direction - most likely other factors at play. Anyway, with Cldc at negative it performs rudder rolls better.

I haven't changed the numerical values themselves in Sabre's rudder and aileron coefficients above - only turned three of them to negative.

And just like with MiG, the handling improved a lot.

Thus, the conclusion:

- yaw damper 0.5

- yaw inertia 0.5 - 2 times of empty weight

- make sure that adverse yaw is present - Cndc of ailerons has opposite sign to Cldc

- make sure that Cydc and Cldc of rudder have the same sign as Cndc

- tweaking coefficients not required!

 

I attach both files for you to try, I am curious if you notice the improvement in handling or it became worse for you instead

Oh, and forgot to add - all tested on Hard FM settings in options only

 

F-100A_DATA.ini

MIG-15BIS_DATA.ini

Edited April 13 by OlWilly

[OlWilly 130]

For the clarity, let's establish what is a rudder and how it works.

Vertical stabilizer is essentially a symmetric airfoil - it doesn't produce lift at zero AoA, aka aircraft flying straight. But as we deflect the rudder, the lift is starting to get produced. Imagine that vertical stab is a vertical wing, and rudder is a flap - this is easier to understand. 

Now, let's say we give right rudder. Vertical stab starts creating lift which pushes the tail to the left. Aircraft "pivots" around its center of gravity, so nose gets pushed to the right. Voila, we achieved a yaw, now longitudinal axis of our aircraft is misaligned to the vector of speed. This is called the yaw angle. 

This creates two effects.

As we now flying "sideways", the airflow over right and left wings is different, and they create different amounts of lift. In our case, right wing creates less lift, left wing more, and aircraft rolls to the right. This is rudder roll.

Then, the relative wind hits the side of our aircraft, in our case the left side. This pushes the aircraft to the right, making it slowly turn right. This is rudder turn.

All of these effects happen on the same side - I.E., as we give right rudder, we get right yaw, right roll and right turn. For aircraft with unorthodox control surfaces like spoilers on B-2 or fully moving vertical stabs on YF-23 the effects are the same. 

After checking more rudder settings on modded and stock planes I found that most of them are messed up. I am not talking the tune-up of coefficients themselves, they simply have the wrong signs.

With the default settings, rudder yaws aircraft correctly, but tries to roll and turn it in the opposite side. This is weird and wrong.

After looking on the forum, I found this thread from 2009 with the guy complaining about the same thing:

https://combatace.com/forums/topic/60218-question-about-flight-model-sideway-g-force/

So the issue is really old, and I can't think of how to explain it.

I can't believe that this is an oversight, in so many planes and over multiple games and expansions packs and patches...

Likewise, I don't think that the devs didn't understand how this works. The coefficients are in place after all.

So the only answer is that this was done deliberately for some reason

[Wrench 9,851]

This is fascinating stuff -- keep it going, and see what needs fixing. I"m sure some of this goes all the way back to SF1

[+mue 715]

On 13.4.2024 at 9:51 AM, OlWilly said:

Cydc is the side force caused by slip. It should be in negative as aircraft will slip in the same direction as rudder applied. Curiously, original file had it positive and aircraft tried to slip in the opposite direction. This is the reason for lack of slip!

Cldc is the rudder roll caused by different airspeeds over inner and outer wings during slip and thus, different amounts of lift. Likewise, should be in negative as aircraft rolls in the same direction as rudder applied, and once again, in original file it was positive.

Here you got it wrong.
Cydc is the coefficient of side force due to control surface (rudder) deflection. E.g. when applying right rudder the rudder generates 'lift' at the tail in the left direction and therefore 'pushes' the aircraft to the left (the original positive Cydc is correct, but maybe too high).
Since the tail is behind and above the cog additionally a rolling moment (counter clock wise) and a yawing moment (tail to the left, or front to the right) is generated. Both moments a defined by Cldc and Cndc.

I think the coefficient you are looking for is side force due to slip angle beta: Cyb. Main contributors are fuselage and tail. You can try to increase this coefficient if you think the side slip force is too low.
 

[OlWilly 130]

 

8 hours ago, mue said:

Here you got it wrong.
Cydc is the coefficient of side force due to control surface (rudder) deflection. E.g. when applying right rudder the rudder generates 'lift' at the tail in the left direction and therefore 'pushes' the aircraft to the left (the original positive Cydc is correct, but maybe too high).
Since the tail is behind and above the cog additionally a rolling moment (counter clock wise) and a yawing moment (tail to the left, or front to the right) is generated. Both moments a defined by Cldc and Cndc.

I concur.

We apply right rudder.

Vertical stab starts producing lift which pushes the tail to the left.

But as aircraft "pivots" around its CoG the nose is getting pushed to the right! This is the yaw - Cndc.

The CoG is the "pivot point" of any such movement. Notice that when you use elevators, the entire aircraft doesn't go up or down - it affects only the pitch, so you have tail going down but nose going up. This is very similar, only on different axis.

The rolling moment is generated by different airflow over both wings. With right yaw, the right wing generates less lift, left wing generates more lift and you get the right hand roll - Cldc.

And lastly, the yaw causes relative wind to hit the side of the aircraft, which turns it right and generates a substantial amount of drag - Cydc . This is why sideslip is often used to lose the energy quickly.

Images attached.

------------------------

But if you still don't agree.

Fire up any other flight simulator. DCS, Lock On, any of Il-2s, MSFS, X-plane, Flightgear, Falcon... Or hop in a real aircraft if you are lucky.

Give hard right rudder.

What do you see?

The plane is yawing and rolling to the right.

Apply left stick to keep the wings level.

What do you see again?

The plane is slowly turning to the right and bleeds energy quickly.

Now do the same in any stock planes in Strike Fighters.

They may rudder roll correctly (depends on the plane), but they won't slip turn at all.

Then apply the changes I described - Cydc, Cldc and Cndc in rudder all negative.

And you will see that the plane starts behaving correctly.

Rudder turning can be used to turn the aircraft without banking - but it is inferior to banking turn because turnrate is much lower and drag increase is profound (also major risk of getting into a spin)

But it is mandatory during air combat when you need to keep sights on the target without banking.

 

8 hours ago, mue said:

I think the coefficient you are looking for is side force due to slip angle beta: Cyb. Main contributors are fuselage and tail. You can try to increase this coefficient if you think the side slip force is too low.

I don't really have to touch other coefficients because the changes I described already do the job.

The one issue I discovered though - the sideslip doesn't cause any additional drag at all. Even at hard sideslip the plane doesn't lose energy at all, you can even go supersonic.

Do you think this could be fixed? I thought about increasing drag of the rudder as a crutch but this simply feels wrong.

Edited April 15 by OlWilly

[+Menrva 4,202]

@OlWilly I may be wrong, but IIRC by setting the flight model difficulty to Easy, the rudder should work more like it should, but the overall flight modelling then becomes Ace Combat-like. Just another test case that is worth trying.

Compared to many other flight sims, the rudder is not useful at all in Strike Fighters, so this thread is of my interest.

[OlWilly 130]

36 minutes ago, Menrva said:

I may be wrong, but IIRC by setting the flight model difficulty to Easy, the rudder should work more like it should, but the overall flight modelling then becomes Ace Combat-like. Just another test case that is worth trying.

Could be, but I really don't see the point in flying on Easy or Normal Flight settings. Even Hard is very forgivable compared to other sims or real life

38 minutes ago, Menrva said:

Compared to many other flight sims, the rudder is not useful at all in Strike Fighters, so this thread is of my interest.

I wouldn't say so.

The game was originally geared to jet gunfighter era, so you have stock planes that did (or were meant to do) a lot of gunfighting, like various MiGs, Super Sabres, Crusaders, French fighters, even Phantoms...

But gunfighting without proper rudder is a torture. You need rudder to gently slip your aircraft left or right without banking to keep the sights where you need them to be.

I can't say that I am particularly good virtual fighter pilot, but I know how to shoot.

But in Strike Fighters, this turned out to be a real embarrassment for me. Even if target flying straight and I have all the time in the world to line up a shot - I still miss. The worst part was when I tailed enemy fighter for 20 (!) minutes, solid on his tail and still couldn't land a good shot.

Yes, SF AI likes to jink very actively, but it does it in very predictable manner: up - down, left - right, force overshoot...

I know that the problem isn't me because I never had such problems in Il-2 or Lock On or DCS...

After fixing the rudder, gunnery became way more pleasant as aircraft does what I expect it to do and now I can enjoy some good jet dogfights

[+Menrva 4,202]

3 hours ago, OlWilly said:

Could be, but I really don't see the point in flying on Easy or Normal Flight settings. Even Hard is very forgivable compared to other sims or real life

I think there's been a misunderstanding. I'm not suggesting to play it on Easy, I never do since I play with Hard difficulty. It's just something worth testing to see how that affects rudders in game.

3 hours ago, OlWilly said:

I wouldn't say so.

Probably my bad wording, I did not mean to say that rudders are not needed in Strike Fighters. What I wanted to say is that they are not useful in the way they have been implemented. Had they been working as expected, they'd have a lot of use indeed.

Edited April 15 by Menrva

[+mue 715]

10 hours ago, OlWilly said:

Notice that when you use elevators, the entire aircraft doesn't go up or down - it affects only the pitch, so you have tail going down but nose going up.

Of course does the elevators contribute to the total lift force. It's basic rigid body physics (keyword: equivalent force systems): e.g. an upward force at the elevator is equivalent to the same force applied to the cog and an additional moment around the cog with moment = force x cog-elevator-distance. Sure, the lift contribution from the elevators is nearly neglectable compared to the main wings, but it's there.

 

10 hours ago, OlWilly said:

The rolling moment is generated by different airflow over both wings. With right yaw, the right wing generates less lift, left wing generates more lift and you get the right hand roll - Cldc.

But that is rolling moment due to slip angle beta Clb (of the wings) and not Cldc (of the rudder).

 

10 hours ago, OlWilly said:

And lastly, the yaw causes relative wind to hit the side of the aircraft, which turns it right and generates a substantial amount of drag - Cydc

That would be drag due to beta Cdb. Unfortunately this coefficient is indeed missing in the SF games.

 

I tested a side slip maneuver with the F-100A: full rudder and opposite aileron to keep the wings level. And indeed the heading doesn't change.
That means that either Cydc (of the rudder) is too big or (the absolute value of negative) Cyb (of the fuselage and tail) are to small, or both. Therefore you can try to decrease Cydc of the rudder (maybe until 0.0) and/or increase the abolute value of negative Cyb (of fuselage and tail).

Edited April 15 by mue

[OlWilly 130]

2 hours ago, mue said:

Of course does the elevators contribute to the total lift force. It's basic rigid body physics (keyword: equivalent force systems): e.g. an upward force at the elevator is equivalent to the same force applied to the cog and an additional moment around the cog with moment = force x cog-elevator-distance. Sure, the lift contribution from the elevators is nearly neglectable compared to the main wings, but it's there.

Yes! But you see, the function of horizontal stab directly depends on the relation between CoL and CoG. 

If CoL is right below the CoG, stab should be a symmetric airfoil and not produce any lift for level flight.

If CoL is in front of CoG, stab should produce the downforce aka negative lift. 

And if CoL is behind the CoG, stab should produce lift. 

But what I say is that Strike Fighters game engine does not take this into account. 

You can move CGPosition= wherever you want without changing a single parameter in horizontal stab, and aircraft will fly just fine. 

F-100 is and old aircraft, CoG is definitely in front of CoL and this means that horizontal stab is rigged for downforce. 

You move CGPosition two meters behind (I just did). What should happen? 

It should pitch up uncontrollably immediately as it gets fast enough. But this doesn't happen. It flies absolutely fine. 

Now let's look at F-16. We know that it has CoG behind the CoL, so horizontal stab produces lift.

But I don't see any fundamental differences between stab parameters of F-100 and F-16. In fact, I just copied stab data from F-16 to F-100. It should pitch down instantly, but it doesn't. 

The game simply doesn't care. It doesn't calculate these forces deep enough.

2 hours ago, mue said:

That means that either Cydc (of the rudder) is too big or (the absolute value of negative) Cyb (of the fuselage and tail) are to small, or both. Therefore you can try to decrease Cydc of the rudder (maybe until 0.0) and/or increase the abolute value of negative Cyb (of fuselage and tail).

Yes, it works.

In fact, we can just purge Cydc and Cldc from the rudder and work with wings, fuselage and tail instead. 

But effect is the same. I think the game treats these values as the same (Clb = Cldc and Cyb = Cydc). 

We can either increase Clb and Cyb values or simply put Cldc and Cydc into negative as "force multipliers".

The latter is simply easier

Edited April 15 by OlWilly

[OlWilly 130]

6 hours ago, Menrva said:

I think there's been a misunderstanding. I'm not suggesting to play it on Easy, I never do since I play with Hard difficulty. It's just something worth testing to see how that affects rudders in game.

Ahh, I get you.

Well, firstly, as we don't have access to source code, I can't really say what the game does when we switch it to easy.

And secondly, I consider the main problem to be mostly solved. Now we have even two methods of dealing with it and with applied changes the rudder does behave like it should. Coefficients could be tweaked around, but this is another story for another day.

6 hours ago, Menrva said:

Probably my bad wording, I did not mean to say that rudders are not needed in Strike Fighters. What I wanted to say is that they are not useful in the way they have been implemented. Had they been working as expected, they'd have a lot of use indeed.

Now they do 

Gunfighting is easier and you can do some tricks with it

Now "Knife Edge" is possible 

 

Hammerhead too

Edited April 15 by OlWilly

[+mue 715]

11 hours ago, OlWilly said:

But what I say is that Strike Fighters game engine does not take this into account

Of course the game engine's FDM does take this into account.

 

11 hours ago, OlWilly said:

You can move CGPosition= wherever you want without changing a single parameter in horizontal stab, and aircraft will fly just fine.

CGPosition= does not do what you think it does.
The cog inside the FDM is fixed at 0,0,0 (in the FDM coordinate system). CGPosition= then positions the FDM in relation to the 3D Model by moving the FDM (cog) to the given position (in 3D model coordinate system).
If you want to change the cog position in relation to the aerodynamic surfaces inside the FDM you have to change the x coordinates of the aerodynamic surfaces instead by altering the Xac*TableData.

 

11 hours ago, OlWilly said:

Now let's look at F-16. We know that it has CoG behind the CoL, so horizontal stab produces lift.

But I don't see any fundamental differences between stab parameters of F-100 and F-16. In fact, I just copied stab data from F-16 to F-100. It should pitch down instantly, but it doesn't.

Since the (real) F-16 is aerodynamic unstable and requires a FBW system, and the game engine does not simulate FBW, I assume that the FDM is adapted to be more aerodynamic stable instead.

[OlWilly 130]

So, with the valuable input from Mue, the updated instructions on rudder related fixes in data.ini (preserve the original file though). Each problem now has two solutions for your choosing

Rudder roll

Pick up the aircraft of your choosing. Give right rudder. Aircraft should gently roll to the right. If this doesn't happen:

Easy way

Find rudder in control surfaces section (two of them for twin tail aircraft). Find Cldc coefficient and put it to negative (add minus sign before the value)

Example: before Cldc=0.0008, after Cldc=-0.0008

More correct way

Find Clb coefficient in wing sections (left wing, right wing, left outer wing, etc). It should be in negative. Increase the absolute value in each of them until aircraft behaves correctly. The increase should be the same for left/right sections, I.E. if you increase the left for 0.05, the right should get the same increase. If this doesn't help or causes weird behavior, decrease Cldc in rudder.

Example: before Clb=-0.0182, after Clb=-0.0232 for both [RightWing] and [LeftWing]

Slip

Pick up the aircraft of your choosing. Give right rudder and left stick to keep the wings level. Aircraft should slowly change the heading clockwise. If this doesn't happen:

Easy way

Find rudder in control surfaces section (two of them for twin tail aircraft). Find Cydc coefficient and put it to negative (add minus sign before the value)

Example: before Cydc=0.1130, after Cydc=-0.1130

More correct way

Find Cyb coefficient in fuselage and tail related sections. It should be in negative. Increase the absolute value in each of them until aircraft behaves correctly. If this doesn't help or causes weird behavior, decrease Cydc in rudder.

Example: before Cyb=-0.5053, after Cyb=-0.5553 for [VertTail]

Excessive swaying on yaw axis

Pick up the aircraft of your choosing. Start giving full right and left rudder in quick succession. If aircraft wiggles like mad:

Easy way

Find EmptyInertia= line. Set the third value to 1-2 times the EmptyMass of the aircraft.

Example: before EmptyInertia=68970.9,20295.1,65332.2, after EmptyInertia=68970.9,20295.1,8000.0

More correct way

Find EmptyInertia= line. Gradually decrease the third value until the behavior becomes acceptable

After all changes

Test the aircraft before and after changes. Set up a simple 1vs1 cannon only dogfight mission, then a simple but cannon only interception mission vs level bombers (Il-28, Canberra, etc), and, finally, do some aerobatics (knife edge, hammerhead, barrel roll with rudder, etc)

 

[OlWilly 130]

More discoveries!

I was playing with rudder drag (CDdc coefficient) to simulate the drag during sideslip. This is really a dirty way of dealing the with issue, but as @mue points out, the game engine simply doesn't calculate the correct force - drag due to beta Cdb.

It works as intended, it is possible to increase the drag so you would start to lose speed during any slip. The byproduct is that you would also experience increased drag during coordinated turns when you apply rudder to correct for adverse yaw, but as adverse yaw in SF flightmodels is not really strong, I can live with it.

But the interesting part is that the value of CDdc also seems reversed.

In all data files I've seen it is set in positive. Makes sense. So, this time I picked Yak-23 as the victim for my tests and started gradually increasing the value. Not only I haven't experienced any drag effect, aircraft actually started to accelerate at full rudder. Then, I went overboard and increased CDdc all the way to 1.0 - and voila, the poor Yak instantly accelerated to supersonic speeds at full rudder. After affixing the value to negative, it started to work as intended.

I went back to my control subject, stock F-100A, and effect is similar. Positive drag value decreases drag and negative value increases it.

This means that every stock and modded plane actually has a reduction of total drag when giving full rudder. It's just that the values are rather low and barely noticeable.

[toot 302]

How do this tweaks effect vtol aircraft/ modded helicopters? If you try and yaw in a harrier whilest in vtol or some other vtol plane it usually starts yawing and then starts struggling with it after a certain point (which might be related to the direction of the aircraft or something, not sure)

[+mue 715]

On 19.4.2024 at 7:10 AM, OlWilly said:

But the interesting part is that the value of CDdc also seems reversed.

...

This means that every stock and modded plane actually has a reduction of total drag when giving full rudder. It's just that the values are rather low and barely noticeable.

Nice find. I could reproduce this bug.
First I feared that all CDdc calculations (e.g. for ailerons, elevator and flaps) were bugged too. But tests showed me that they work correctly (i.e. the drag increases and the aircraft slows when these control surfaces are deflected).
It seems that ReverseInput=True in the Rudder section triggers the bug (the ReverseInput option is absent in the other control surface sections).

 

[OlWilly 130]

On 4/19/2024 at 11:22 AM, toot said:

How do this tweaks effect vtol aircraft/ modded helicopters? If you try and yaw in a harrier whilest in vtol or some other vtol plane it usually starts yawing and then starts struggling with it after a certain point (which might be related to the direction of the aircraft or something, not sure)

Control of such aircraft in VTOL mode is taken care by different control inputs. The max control speed for rudder and ailerons in Harrier is 200km/h which means that below this speed they do nothing.

The maneuvering below these speeds is governed by engines themselves which is how the game simulates differential thrust I assume. 

This is the correct approach but the game doesn't really like any aircraft staying in the air at airspeeds close to zero. Every time I tried to make any VTOL to hover it started to show the stall behavior (which is not correct, the lack of lift is compensated by engine thrust) and just fumbled uncontrollably