Useful Landing Gear Adjustment

[AdmiralTigerclaw]

Just finished a flight test on some tweaks I've been doing for my own game. 

 

I'm willing to bet many of you playing with aircraft have problems with skidding and spinouts from even the slightest yaw input while on a takeoff roll or landing.  Most people don't make sense of the new tweakables, and settings, and I've noticed that some critical settings are counter-intuitive.  I have some workaround fixes that may improve ground roll performance.

 

I'll start by advising you, part of this light fix will be modification to Squad parts files.  Back them up.

And here's the fix list.

1: Input Sensitivity value is inverted.   Set HIGH, not LOW.

In your game settings, under input for vehicles, you'll have a sensitivity slider that ranges from 0 to 100.  For some reason, the value mode is inverted.  0 is HYPER sensitive  (The slightest tap takes the control range to max).  Set your sliders to 100 to minimize sensitivity.  This is likely a bug, so be prepared to verify this on any given update of KSP.

2: Use increased, but not identical friction on gear.

When setting up landing gear in tricycle arrangement, set the friction value of the aft gear to 2, and the friction of the steering (forward) gear to 1.5.  In this manner, your forward gear will slip while your aft gear are still gripping.  So just like tires on cars:  Newest tires with the best grip go on the back to prevent the aft end coming free.  The increased slippage on the front gear also prevents an aggressive turn command on the steering from grabbing too suddenly.

 

3: Modify the Steering Curve.

While fixes one and two make craft somewhat controllable, they still get extremely squirrely if given the slightest direct yaw steering input.  The result is a bunch of sensitive overcorrection taps that make things ugly VERY fast.  I did some digging, and brought up the module that defines steering behavior.

MODULE
	{
		name = ModuleWheelSteering
		baseModuleIndex = 0
		
		caliperTransformName = SteeringPivot
		
		steeringResponse = 2		
		steeringCurve
		{
			key = 0 30
			key = 10 9
			key = 30 2
		}
	}

This is code for the small retractable landing gear  (Single wheel).  Immediately, I noted 'steeringResponse = 2' and 'steeringCurve'.   I was unable to find anything on steeringResponse, but I found some discussion around the forums talking about the values under Steering Curve.  Without any final say, the conjecture is that the first value is velocity, the second is turning angle, I'm assuming in degrees per second.  Which is saying in the code above that that the rate the wheel can turn is 30 degrees per second at 0 m/s up to the maximum pivot value.

It is ALSO possible that that is the maximum pivot value for the wheel at said speeds. 

Either way, I halved the first number and left the second alone.  If it's a rate curve, then the rate of wheel pivot decreases with speed, making it react slower as you gain speed down the runway, decreasing the overcorrection tendencies.  If it's a maximum angle, then you STILL benefit by the reduced velocity value as the maximum turn angle is cut down much sooner.

 

So, long and short, go down the keys and half the value of the first number in each key.  This should improve turning control of aircraft on flight rolls.  I noticed an immediate improvement when I did so.

 

 

As a note for anyone who wants to play with modifying these values:

10 meters a second velocity is 19.4 knots, which is 22 MPH, which is 35 km/h.  Keep those speeds and values in consideration when deciding how much turn you want wheels to give you.  Anyone who learns more, feel free to add to the topic.

 

 

Edited May 13, 2016 by AdmiralTigerclaw

[jlcarneiro]

Thanks! I'll try them.

A suggestion, though: why don't you propose a patch using Module Manager (MM)?

I THINK it should look somewhat like this:

@PART[SmallGearBay]
{
    @MODULE[ModuleWheelSteering]
    {
        @steeringCurve
        {
            @key = 0 30
            @key = 5 9
            @key = 15 2
        }
    }
}

I'm not that experienced with it, so take this syntax is PROBABLY WRONG! Take it with extreme care...

[AdmiralTigerclaw]

I'd say let this run its course a little first.  A few more confirmations that this is working and some tweaking to see if other values do better would be nice.

[Alshain]

One of the issues with steering is in the past in 1.0.5 or earlier it would cut off after the craft reached a certain speed so you could only use yaw.  In 1.1 it does not, steering is active at all speeds, and to make matters worse, by default Yaw and Steering are assigned to the same controls.  So you attempt to yaw and get steering too.  I believe this to be a bug personally, but Squad probably sees it as a feature request.  You can mitigate this by reassigning your steering keys to something that isn't Yaw also.

Edited May 13, 2016 by Alshain

[AdmiralTigerclaw]

Just now, Alshain said:

One of the issues with steering is in the past in 1.0.5 or earlier it would cut off after the craft reached a certain speed so you could only use yaw.  In 1.1 it does not, steering is active at all speeds, and to make matters worse, by default Yaw and Steering are assigned to the same controls.  So you attempt to yaw and get steering too.  I believe this to be a bug personally, but Squad probably sees it as a feature request.  You can mitigate this by reassigning your steering keys to something that isn't Yaw also.

I use a stick, so I actually WANT the steering applied along side the yaw control.  Technically, that's how it is on actual aircraft.  Rudder pedals control the rudders on the V-stab, but also the Nosewheel Steering mechanism.

 

What you/we want is for the gear turn sensitivity to decrease as velocity increases.  That way, a tap that would give you a sharp taxi turn on the ground doesn't translate into cartwheeling when correcting minor nose drift late in a takeoff roll.

 

I'm going to try changing the steering response down.  It strikes me as odd the value is 2, and I believe that might be a scalar...  If it IS a scalar, that means it's doubling all values plugged in.

 

 

[Alshain]

4 minutes ago, AdmiralTigerclaw said:

I use a stick, so I actually WANT the steering applied along side the yaw control.  Technically, that's how it is on actual aircraft.  Rudder pedals control the rudders on the V-stab, but also the Nosewheel Steering mechanism.

 

What you/we want is for the gear turn sensitivity to decrease as velocity increases.  That way, a tap that would give you a sharp taxi turn on the ground doesn't translate into cartwheeling when correcting minor nose drift late in a takeoff roll.

No, the issue is more or less having to use yaw at high speed to correct your plane on the runway because the wheels in 1.1 veer off every which way.    Actual aircraft wouldn't have to make huge corrections just to stay on the runway like we do in 1.1.  In 1.0.5 steering cut off COMPLETELY at around 35-40 m/s, you could only use yaw after that.  In 1.1 it does not, so while using yaw to stay on the runway, you are also using steering and that is way too sensitive.

Edited May 13, 2016 by Alshain

[AdmiralTigerclaw]

12 minutes ago, Alshain said:

No, the issue is more or less having to use yaw at high speed to correct your plane on the runway because the wheels in 1.1 veer off every which way.    Actual aircraft wouldn't have to make huge corrections just to stay on the runway like we do in 1.1.  It 1.0.5 steering cut off COMPLETELY at around 35-40 m/s, you could only use yaw after that.

I belive that's what:

		steeringCurve
		{
			key = 0 30
			key = 10 9
			key = 30 2
		}
	}

 

Is supposed to be doing, but something is wrong with the default settings. 

 

If the first value is m/s, and the second is angle, then we have a curve that drops from 30 degrees to 9 degrees from zero to ten meters a second, and then 9 to 2 from 10 m/s through to 30.  And if steeringResponse is a scalar override like I supect, it's effed up because that all gets doubled.

 

Back in a bit, going to test.

 

EDIT:

Okay, the steeringResponse is the time response scalar.  Setting it to 0.2 for an extreme low resulted in a gear turn rate that was abysmally slow.

 

PROGRESS!

Edited May 13, 2016 by AdmiralTigerclaw

[KerbonautInTraining]

Btw, a lot of real aircraft don't use the same steering controls during taxi and takeoff.

Typically something like an airliner can steer it's nose wheel a few degrees with the rudder and has a separate lever for large steering motions. 

I think I'm gonna set it up so my pedals control the rudder and twisting the stick does wheel steering. But seeing as most people don't have such a control setup, the best solution IMO is to have an action group to disable the steering during takeoff. 

[Alshain]

7 minutes ago, KerbonautInTraining said:

Btw, a lot of real aircraft don't use the same steering controls during taxi and takeoff.

Typically something like an airliner can steer it's nose wheel a few degrees with the rudder and has a separate lever for large steering motions. 

I think I'm gonna set it up so my pedals control the rudder and twisting the stick does wheel steering. But seeing as most people don't have such a control setup, the best solution IMO is to have an action group to disable the steering during takeoff. 

I don't have pedals, but I have a similar setup.  My throttle has an analog on the back of it that use for steering, and then I use the stick for yaw.

[AdmiralTigerclaw]

Alright, new values:

 

steeringResponse = 4

curve

key = 0 40

key = 5 5

key = 15 1

 

The yaw command is very snappy with a response of 4 which prevents the boat-like correction lag I found at lower settings.  I'm kind of liking it at four.  It responds nicely to input.  I'm going to try messing about with the curve and see about getting it to play nice on takeoff rolls better.  Right now it still seems overly sensitive.

[MaxPeck]

This is promising!

I agree with @jlcarneiro that this would be better run in a MM patch though - you wouldn't have to mess with the original file, and it's much easier to make corrections, because you can try different values and then I think you can F11 to reload them without having to end and restart your game.

[Shadriss]

So let's assume for a moment that this is doable through MM. What would that patch look like to account for ALL wheels? Also, and I suppose this should have been the first question - will you release this as an MM config until either A) Squad gets wheels figured out or B) Adjustable Landing Gear gets updated (which may very well be after A... in fact, likely).

[gilflo]

What I get on 1.1.2 is for wheels a primary sensitivity from 0.01 to 1 and a secondary sensitivity from 0.01 to 1 the initial setting  is that the slider is indicating 1, so I think it's the setting you recommend !

Do you set the same values  you recommend for  steering response and key curve for all types of gear?

[AdmiralTigerclaw]

35 minutes ago, gilflo said:

What I get on 1.1.2 is for wheels a primary sensitivity from 0.01 to 1 and a secondary sensitivity from 0.01 to 1 the initial setting  is that the slider is indicating 1, so I think it's the setting you recommend !

Do you set the same values  you recommend for  steering response and key curve for all types of gear?

 

I set the steering response and key curve the same for all gear, yes.   At least, the gear that have it.  The large and XL gears seem to be intended to be large fixed beasts.  But yes, same response curves.  I don't see why one would change the steering curves for a larger gear when you want the same rotational tracking behaviors.  I want the nose gear to have limited response as the vehicle gets up to speed.

 

I may also test out maxing out friction later this evening and seeing how some new values in my steering response affect aircraft spin tendencies.  I'm controllable, but I still have to use very light taps for steering corrections.

Edited May 24, 2016 by AdmiralTigerclaw

[gilflo]

ok thank's for the reply

[problemecium]

I choose option 3: No, because I'm fine with wheels the way they are already. No offense meant to anyone, but it is missing from the poll ;P

[gilflo]

Here are my setting and observations on landing with a 53T SSTO:

I use the steering response and curve from AdmiralTigerclaw

Rear gears are LY 35 medium landing gears and settings are: Friction 2 - Spring 0.6 - Damper 1 - Brakes 200 - Steering disable

Front gear is also a LY35 medium gear with steering enable: Friction 1.5  Spring 0.6 - Damper 1 - Brakes 150

For what i tried landing on grass near the runway can be done without damage at a vertical speed of 8m/s with no bouncing, no skidding. Just before touch I reduce to iddle and apply full brakes by pressing B, but even with 2 speed brakes and those settings braking run is long.

I notice that on full power with the parking brake set (hit the toggle brake icon on the top center of the screen), the plane is skidding.

Landing on runway with the same settings is done without bouncing, but if you loose steering control the plane skid head to tail and stop quickly......

Steering control is still an issue on landing and is very touchy and landing with no bouncing seems easy with those settings....but landing run seems very long and brakes not so efficient at max settings

[gilflo]

I think you can take off and land any aircraft provided you got enough gear to support the weight and the good landing gear settings.

For landing it's a bit difficult to stay on the runway, landing aside is easier. Here is a 478T SSTO supertanker with stock gears, 28 Wheels, no problem to land on the grass but very long run to stop it

even with max brakes on 20 rear wheels (200), 150 on 8 front wheels and 2 drag chutes. Friction is 3.5 on the rear and 2.5 in front. All Springs at 0.6. Vertical landing speed around 5-6 m/s and landing speed 110m/s, no bouncing, but not on runway..every time I tried it looks like the runway is icy! grass is far better!!

This aircraft being very heavy with big inertia, Pilot assistant helps a lot.

[The_Rocketeer]

@gilflo you're giving me deja-vu man!

[Tfin]

On 6/5/2016 at 7:11 PM, gilflo said:

I think you can take off and land any aircraft provided you got enough gear to support the weight and the good landing gear settings.

All of which is just working around the broken gear. I've been able to land using the early landing gear by tweakscaling them to 150%, but the impact they can handle is nowhere near what the stats on them in the hangar indicate. A minor bump blows them up as I'm trying to drive science experiments around KSC. I'm afraid to send anyone to the Minmus ground survey contract sites I have since there have been reports that legs are affected too.

[gilflo]

I revert steeringResponse to 2 and increase friction on nose gear to be more important than on main gear. This way the nose gear response is better and if you skid head to tail you turn around the nose gear without damage. I also try take off with no steering on nose gear and i think it's easier, you just steer with the rudder and it's efficient enough, but not satisfactory enough in the way that it can be still to powerfull.

Edited June 9, 2016 by gilflo

[gilflo]

I confirm my settings are far better to landing take my big SSTO on runway.

[Araym]

Solution to my problem during landing with "new" landing gears from 1.1.2/1.1.3:

1- as EVERY KSP version, reassigned "steering" for wheels to the same keybinding for yaw in planes with vertical rudder(s): command need to be correlated...

2- ... then, as I preferred "older version", almost removing all "spring" value (... rarely I use above 0.3) while increasing "dampner" MINIMUM at 1.5: landing gears tend to settle any plane lower (as they sink on their own weight), but regain the old appeal of "solid landing gear", almost removing the bounciness of stock values: I came very shallow on approach landing smoothly, settling on brakes pretty well (Long story: low spring value to remove bounciness, hi dampner value to remove movements up and down, basically returning to the feeling of a "fixed landing strut with wheels" rather than an "ammortized one")

3. NEVER used the fixed gears: they are broken!

I landed pretty everywhere on Kerbin with this setup, allowing them also some long taxing in case of missions requiring multiple ground samples around.
The very same set up used on up/downscaled version elaborated maybe in some design using tweakscale mod.

[gilflo]

My solution on 1.1.3

-a stable plane in flight will be stable on ground

-No need to modify steering curve

-Only enable steering on front gears.

-All spring to 0.7 and Damper to 1

-Rear and main gear friction to 1.5 and front gear to 1.0 - You can even disable steering on front gears for take off and landing, rudder control should be enough to steer on take off course.

-Main gear near COG and don't hesitate to add belly gears on heavy belly craft.

-Tweak your gear if necessary to support your plane weight

-Use this mod to have all your gears perfectly aligned in 3 dimensions

-Brakes always stronger on main and rear gears

-If gears can support the weight of your plane, the plane should not dodge on launch and the wheels should not be stressed.

-Well dimensioned gears can land a plane without flare at a vertical speed of 10m/s

I got a 470T SSTO with 2 6 wheels main gears (130% tweak) behind COG and 2 6wheels belly gears (100%) around COG and 2 front medium landing gears (tweak 175%) built this way.

[Rocket In My Pocket]

Wheres option #3?

"I haven't experienced any issues with takeoffs/landings."

Edited July 18, 2016 by Rocket In My Pocket