Car's physics and setting...

[DonaemouS]

Can I propose, as I did on the ASW Forum too, to start a thread in which we can talk about "general" .car tweaks and standardization?

I really interested in how big brains as NNB and Tomte (just to say a couple) writing to tweaks any value inside the .car or just how to get the best for any specific car. But, what is painful is searching about these infos, one months later.

So, in IRC, Tomte was really kind to explain me how the:

Code:

wheels.loadSensitivity

work, and how to set this value to have a better simulation experience. Yesterday, with Tomte, we would also start to focus on:

Code:

suspensionFriction

Well, I'm ready to read your comments. In first, what is it? Is it the value of how many "strenght" the suspensions lost while turning on the road?

I mean, turning, is up to tires, ok, but suspension get a lateral force too to fight, trying to keep the tires on the road, avoiding small jumps (due to G forces).

[Tomte]

jonas wrote about supsensionfriction:

This value is a constant damping factor for suspension movement. Unlike the damperStrength, it is unrelated to how fast the shocks move (since more oil/gas will be compressed, shocks will damp more when you try to move them faster).

I understand it as follows:
supsensionFriction and damperStrength both dampen the movement of the wheel/car body.
supsensionFriction is a base damping (constant), much like engineBaseFriction.
damperStrength is a factor which allows for damping dependant on the speed the wheel/car body is moving up and down. The faster the wheel moves up or down, the more damping is applied. This would be similar to engineRPMFriction.

So we have 2 damping factors, which together make up the damping of the car.

About spring rates, jonas wrote the following:

The spring rates are calculated based on car weight, so two cars with different weights but the same suspension travel will always sit at the same height.

That means that Redline calculates the spring rates. Therefore, we cannot specify a stiffer springs directly, (and here comes my interpretation:) only by lowering the suspension travel. Lower suspension travel would mean higher spring rates and vice versa.
I guess sometimes it would be useful to have a high spring rate AND high suspension travel, for example if you have a quite heavy saloon.
To a certain extend, we can make the suspension stiffer by raising both the supsensionFriction and damperStrength, up the the point when the car is over-damped and cannot follow the bumps in the road anymore.
The inertia values (especially X and Z) can make the car feel stiffer, but that's not the case. I think the inertia values only delay the load transfer when cornering or accelerating (+/-). The load transfer will still be the same once the suspension settled.

[DonaemouS]

Here an extract of our IRC morning chat about wheels.loadSensitivity:

Code:

08:20 DonaemouS|tRc i need to understand  better how load shift works
08:21 DonaemouS|tRc i always thought load shift was the tire pressure (referring to wheels.loadSensitivity)
08:21 Tomtar0 no, load shift is often called weight shift.
08:21 Tomtar0 but the weight does not shift that much, it's the load of the tires
08:21 DonaemouS|tRc need i to search on wiki?
08:21 Tomtar0 not necesarily
08:22 Tomtar0 load shift is what happens when you brake or accelerate
08:22 Tomtar0 when you accelerate, the load (i.e. mass of the car) gets shifted to the rear wheels, ok?
08:22 Tomtar0 hence you have better traction with a RWD car
08:23 Tomtar0 and worse with a FDW car
08:23 DonaemouS|tRc right
08:23 Tomtar0 under braking it is the opposite: the front wheels get more grip, the rear wheels get 'light'. You know that
08:23 Tomtar0 that's why we have more brake force at the front than at the rear
08:23 DonaemouS|tRc but, you wrote "is what happens" to what?
08:24 DonaemouS|tRc i mean, is what happens to body, to suspension, to tires?
08:24 Tomtar0 yes
08:24 DonaemouS|tRc which one?
08:24 Tomtar0 well, they are all conndted
08:24 Tomtar0 ok, the body shifts relativly to the tires, if you will
08:24 DonaemouS|tRc ok, but load shift is a different variable on each wheel
08:25 Tomtar0 no it isn't. The load shift is unaffected
08:25 DonaemouS|tRc im missed myself
08:25 Tomtar0 the sensitivity to the loadshift can be controlled via loadsensitivity
08:26 Tomtar0 this just tells the physics to lessen or highten the effect of the loadshift under acceleration/braking/cornering
08:26 DonaemouS|tRc lower is -?
08:26 Tomtar0 yes
08:26 DonaemouS|tRc so
08:26 DonaemouS|tRc if i use a -0,07 on front wheel
08:26 DonaemouS|tRc this mean, under accelerating
08:27 DonaemouS|tRc the front wheels lost less grip?
08:27 Tomtar0 exactly
08:27 DonaemouS|tRc a positive value it mean they get grip?
08:28 Tomtar0 + means the wheels are more affected by the load shift
08:28 Tomtar0 meaning less grip
08:28 DonaemouS|tRc ok ok
08:29 DonaemouS|tRc if i use 0 is normal lost, plus they lost much more traction, minus they suspension apply resistances and lost less grip
08:29 Tomtar0 yes
08:29 DonaemouS|tRc on rear?
08:29 Tomtar0 same effect here
08:30 Tomtar0 but under accelerating a negative value means that you don't have as much grip as you could have but under braking it means you have more than you could have with 0 or a positive value
08:31 DonaemouS|tRc so if i use -0,05 on front i should use 0,05 on rear to compensate weight loading or no?
08:31 DonaemouS|tRc the sum should be 0 or not?
08:31 Tomtar0 the sum dies not have to be 0
08:32 Tomtar0 if it is not 0, some weight is just magically lost
08:32 DonaemouS|tRc now help me to understand an heavy sedan rwd
08:33 DonaemouS|tRc front must be a minus something
08:33 Tomtar0 why?
08:33 DonaemouS|tRc cause the suspension need to help directionality otherwise, while accelerating the car cant turn
08:33 Tomtar0 that means more understeer
08:33 DonaemouS|tRc ok
08:34 Tomtar0 if you use a - value in front you get:
08:34 Tomtar0 less weight->grip->traction when braking in front and a bit more traction when accelerating
08:35 Tomtar0 that means more understeer when braking and less when accelerating so I would not necessarily touch the front
08:35 DonaemouS|tRc for front wheels
08:35 Tomtar0 leave it to 0
08:36 Tomtar0 the rear is the important bit
08:36 DonaemouS|tRc lets see
08:36 Tomtar0 I always aim for stability under braking
08:36 DonaemouS|tRc and traction?
08:36 Tomtar0 I don't want the rear wheels to loose traction too easily when braking
08:36 DonaemouS|tRc but
08:37 Tomtar0 so I use - values
08:37 DonaemouS|tRc a sedan is a front engine, front weight front all except for trasmission
08:38 Tomtar0 ok, a negative value will give you less traction when accelerating
08:38 DonaemouS|tRc i supposing, cause the weight maybe not transfer on rear
08:39 Tomtar0 but that is not too bad because most of our cars accelerate too quick anyway
08:39 DonaemouS|tRc the car should move the engine block weight on rear
08:39 Tomtar0 kinda
08:39 Tomtar0 but
08:39 Tomtar0 under braking, you have less weight shifted to the front
08:39 DonaemouS|tRc ummh is this so bad? i mean what of the rear weight should be moved on front?
08:40 Tomtar0 the weight gets moved nevertheless the weight always moves, the question is how it does affect the wheels
08:40 DonaemouS|tRc indeed
08:41 Tomtar0 with a negative value, the wheels are affected less
08:41 DonaemouS|tRc so load is much more related to suspension then wheels
08:41 Tomtar0 well, both
08:41 Tomtar0 first suspension, then wheels, but in the end always the wheels even with a super stiff suspension or none at all, you still have load shift, but with no suspension you don't have weight shift.
08:47 DonaemouS|tRc yes cause as we wrote engine block should not move so easily on rear at least not in a 1,9 tons sedans, also, just for pure fun i love the car spinning on acceleration
08:49 Tomtar0 the loadsensitivity is a bit of a cheat, but is the only way to set up the rear suspension different to the front

[C14ru5]

Very good info so far.

Redline calculates the spring rates. Therefore, we cannot specify a stiffer springs directly, (and here comes my interpretation:) only by lowering the suspension travel. Lower suspension travel would mean higher spring rates and vice versa.

I don't know what you mean by "stiffer springs" - are you talking about the resonant frequency of the suspension, or do you mean the amount of travel the springs do after a certain impact (say a 1m freefall drop)? If I remember my physics from school correctly, those parameters depend on each other (F=ma, etc.). At least to me it seems like in addition to wheels.maxSuspension, supsensionFriction does something to the spring stiffness (by that I mean the amount of "freedom of movement" the springs have compared to the car body). Try driving over bumps with different supsensionFriction settings with zero damperStrength, and the resonant frequency of the car suspension does change depending on the supsensionFriction. Perhaps we are both thinking in the same way, just that we use our terms differently.

Code:

differentialLockCoefficient

It is just a coefficient regulating how much torque will be transfered from one wheel to the other if the velocitys don't match -- the calculation is mass independant, so you would have to use higher values for heavier cars to get the same effect

it will always transfer torque from the faster to the slower wheel, just the amount transferred is dependant on the difference (and the coefficient of course)

This setting is a hard one to get right, often because it is difficult to test. Many handling problems in suspension (travel/swaybars), engineFriction or wheels.inertia are wrongly interpreted as being caused by the amount of differential locking. So far, there is no way to safely say how much differential locking (say 50% LSD) that occurs in a Redline car - this remains a setting that must be tested through trial and error for each car you make.

The easiest way I've found of telling if you have too little or too much differentialLockCoefficient with RWD cars: Start off with the coefficient at zero, and test the car's responsiveness during oversteer correction (like driving through a high speed chicane). Too little and the car will more or less turn in by itself, but then carve its way through the turn in a sluggish oversteer motion. Too much and the car will feel stiff on corner entry, and it will try to straighten itself up really quickly, and jerk/shimmy itself back into balance. Some of these effects wouldn't happen with a real life LSD (they are typically more related to load transfer issues), but this is how it seems to work inside Redline.

[Tomte]

stiffer springs

I mean springs with a higher spring constant, i.e. a larger force opposing the tension/pressure brought into the spring. F=ma, with F=-kx (k = spring constant, x = change in length from resting position). The only thing we don't set directly is k, with x, m and a sort of known. k is calculated by Redline to keep the car in it's resting position when no external forces are applied on the body (apart from its weight obviously).

If, as jonas stated, supsensionFriction is a damping factor that is independant of the speed the spring/shock is compressed, then surely it will appear like a stiffer spring, but it is actually just a linear friction. At least this is how I understand it.

[Tomte]

I'd like to know what te maxClutchTorqueTransfer is for.

Is it just the obvious, that we could limit the amount of torque a clutch is able to transfer? Say we have an engine which produces 1200 Nm but the clutch can only hold 900?
If this is the case, we might be able to simulate the very flat torque curves of modern turbo charged engines (which have for example a max torque of 310 Nm from 2200 to 5000 rpm). But this would need some careful massaging of the required numbers.

Or is this value some torque limit further down the line in the drivetrain, i.e. after the gearbox or after the final drive? Just to clarify: which every reduction in rpm from the gearbox or the final drive, the torque raises. Otherwise you couldn't move a car with just say 150Nm.

Maybe some of the more knowledgeable people with an insight to how it's applied can chime in here.

[C14ru5]

Tomte, you're the torque expert here, do you honestly not want to try to answer this one yourself? No problem, I'll give it a try from what I've experienced.

Code:

maxClutchTorqueTransfer

This value should be valid for any RPM range. I've seen three uses:

1. It may limit the maximum torque output when power is applied. Redline uses a cubic root interpolation in order to calculate the torque curve, using the value specified at torqueRPM and the value it calculates from the power output at powerRPM. If the relationship between these values causes the cubic root interpolation to give a wrong torque curve result (like thomm's McLaren M8B - i told him those values were wrong...), you may use maxClutchTorqueTransfer to eliminate any unwanted peaks in Redline's interpolated torque curve. The other similar use for this value is to apply a very high torque value at low RPM, and dramatically flatten the torque curve with maxClutchTorqueTransfer to either simulate turbos or simulate the flat torque curve of an electric motor.
   
2. Severe downshifting which would ruin the gearbox in a normal car. Ever wonder why you in some cars can force downshift instead of brake, without losing control? A rather low setting here is why. Realistic or unrealistic? Discuss if you must. Personally I don't think normal drivetrains can cope with 900 Nm, but this must be weighed against downshifting behavior, I guess.
   
3. Jerk/glide behavior during standing starts. In my karts (not updated with new clutchRPM yet, sorry), the standing start behavior is clearly visible. If maxClutchTorqueTransfer is much higher than the kart's maximum torque output, the kart will jerk off the starting line instead of slowly gliding up to speed using the clutch. EDIT: This effect is only noticable if your car doesn't produce wheelspin during standing starts. It may also be an effect from a torque curve estimate that perhaps wraps past 0 Nm and into high values when approaching 0 RPM, I don't know.
   

[DonaemouS]

Very good info so far.
I don't know what you mean by "stiffer springs" - are you talking about the resonant frequency of the suspension, or do you mean the amount of travel the springs do after a certain impact (say a 1m freefall drop)? Try driving over bumps with different supsensionFriction settings with zero damperStrength, and the resonant frequency of the car suspension does change depending on the supsensionFriction.

Ok, we now know what is it, but how is calculated remain a mystery. I mean. Braking in in Nm, sizes in meters, power in Kw and so on.

I have

Code:

suspensionFriction 1100

in almost all my cars. I leave DANDY's BMW M5 setting. But this 1100 what mean? What is a 0 (zero) value? what happens rising from "zero" or reducing?

[Tomte]

Tomte, you're the torque expert here, do you honestly not want to try to answer this one yourself?

I know how the torque is calculated over the rev range in Redline, somebody provided me the formula. But that's not the question.
Down the drivetrain, the torque the engine produces is multiplied several times by the gear ratios, the final drive ratio and finally the wheel radius. So what you get at the wheel is (normally) more than what you get at the engine. Hence my question where the maxClutchTorqueTransfer is applied.

Code:

maxClutchTorqueTransfer

[*]... dramatically flatten the torque curve with maxClutchTorqueTransfer to either simulate turbos or simulate the flat torque curve of an electric motor.

I fully agree here. But I already see people moan that a car shows a too high torque value on the car selection screen.

[*]Severe downshifting which would ruin the gearbox in a normal car.

So a low value allows for downshifting without losing control? Or the other way round? I could also check myself...

[*]Jerk/glide behavior during standing starts.

Gotta check that as well.

It may also be an effect from a torque curve estimate that perhaps wraps past 0 Nm and into high values when approaching 0 RPM, I don't know.

As far as I know Redline does not calculate abnormal high values at or near 0 rpm. From 0 to torqueRPM, it's a simple square raise form 0 Nm to maxTorque.

[NoNameBrand]

As far as I know Redline does not calculate abnormal high values at or near 0 rpm. From 0 to torqueRPM, it's a simple square raise form 0 Nm to maxTorque.

That's too bad. The ZR1 has at least 500lbs-ft of torque from 2000 RPM.... Redline will only give it a paltry 430lbs-ft at 2000 RPM ...

[DANDY]

Weighing in....

If maxClutchTorqueTransfer is much higher than the kart's maximum torque output, the kart will jerk off the starting line instead of slowly gliding up to speed using the clutch.

This tells me the setting reduces the effect of the clutch. Going higher than max torque probably has the same effect as installing a performance or racing clutch, while leaving it the same might be closer to a factory clutch. Assuming this is correct, setting it to less than maximum might simulate drivetrain power loss.
-----------------------------------

Code:

 suspensionFriction 1100

I'm guessing that's kg/xm ....or kg(of resistance)/[unknown distance unit of travel]
-------------------------------------
SWAYBARS:
I've been guessing at the swaybar thickness in mm and multiplying by 1000. It's crude, but typically got me the results I wanted.
Well, I found a spec on the XBOW: Torsional resistance=36000 Nm/degree. Could this be what we're tuning?
Does this mean body stiffness only, or does it include the effect of swaybars?
I've made some guesses based on that figure for some other cars, and I like the results.

[Tomte]

Well, I found a spec on the XBOW: Torsional resistance=36000 Nm/degree. Could this be what we're tuning?
Does this mean body stiffness only, or does it include the effect of swaybars?
I've made some guesses based on that figure for some other cars, and I like the results.

I think this is only the stiffness of the body, not including suspension 'stiffness'.
It just means that if you would fix one end of the body and twist the other end along the z axis, you would need 36000 Nm (or about 3.6 metric tons pushing on at the end of an 1m long lever) to get 1 degree difference between the front and the back. I assume that the fix points would be the axles (with a rigid 'suspension'). I guess that this is rather stiff.

About your method to use the swaybar thickness x 1000 to get a swaybar value: I think you were lucky that the numbers fitted so well on the 911. You're right to assume that the diameter of the swaybar will influence the torsional stiffness to a great deal (apart form the chosen material obviously).
The swaybars are (afaik) connected to the wheels using a link. Swaybars get twisted when one wheel has a different height than the opposing wheel, compared to the body. If this link has not the same length at the front and rear, the leverage to twist the torsion beam/swaybar is different. No idea if the length of these links is known, but I guess that is rather difficult to find out.
Hence I don't see this as a general approach to find out good swaybar settings.

[DANDY]

Code:

wheels.loadSensitivity

Be careful with this stuff, it's very strong, and quickly deadens the handling.
Changing from .02 to .01 (front, rear set lower and changed in relation) increased my TCC BMW WIP's Black Forest lap times by 2 seconds. It basically removes the challenge of driving fast in percentages (0.1=10% easier, in non-scientific, seat-of-the-pants type measurements)

About your method to use the swaybar thickness x 1000 to get a swaybar value: I think you were lucky that the numbers fitted so well on the 911. You're right to assume that the diameter of the swaybar will influence the torsional stiffness to a great deal (apart form the chosen material obviously).

...and it works fine on every other car I've made, provided my guesses as to thickness are accurate (I don't think they all are). In trying to find a standard (if only for myself), this one works well, even though I've always known the link between thickness and strength isn't so well-defined.

The swaybars are (afaik) connected to the wheels using a link.

They also connect to the body.

Hence I don't see this as a general approach to find out good swaybar settings.

Then what is a good approach? Fiddling till you like it? They have a significant effect on handling, and various plug-makers' settings are all over the map. A heavier car will obviously need a stronger setting to get the same effect as a lighter version, but where does one stop? When their vintage sportscar or touring car holds the road like a modern formula racer?

[Tomte]

...and it works fine on every other car I've made, provided my guesses as to thickness are accurate (I don't think they all are). In trying to find a standard (if only for myself), this one works well, even though I've always known the link between thickness and strength isn't so well-defined.

I don't deny that. All I'm saying is that I wouldn't use the diameter of a swaybar as a general rule to be applied without thought. It might work very well as a guideline if you know what parts make a suspension. I don't critisize your approach, I'm just warning other plug-in makers that it doesn't always work.
I very much like he handling of your cars and I think they are one of the most realistic drives around, with the 911 simply astonishing accurate (at least to how I imagine the car to behave).

The swaybars are (afaik) connected to the wheels using a link.

They also connect to the body.

Only if you mean the bushings. Normal, nonadjustable swaybars are not connected to the body in a sense that they push against it. They only interconnect the wheels on the same axle. Otherwise it would be a torsion beam suspension (like the Volkswagen Beetle). See swaybar entry on Wikipedia

Hence I don't see this as a general approach to find out good swaybar settings.

Then what is a good approach? Fiddling till you like it?

Basically, yes. And a bit of luck. And comparing with other, hopefully similar redplugs.
I mean, the Atom doesn't even have swaybars in real life (apart if one adds them oneselves). Does that mean I should leave them out of the .car? Of course not, because swaybars are the main influence on the handling balance of our cars. The Atom would be undrivable without.
There are so many suspension settings influencing the handling balance of a car that we cannot accomodate in Redline, like geometry, type, camber, toe in and so on, so we need to play with what we have.

[DANDY]

I mean, the Atom doesn't even have swaybars in real life (apart if one adds them oneselves). Does that mean I should leave them out of the .car? Of course not, because swaybars are the main influence on the handling balance of our cars. The Atom would be undrivable without.

Sounds like a challenge.

[DANDY]

Sounds like a challenge.

Raised sups.friction to 300, stiffness to 12000, set diff to 90 and pulled off the swaybars. It's quick and dirty, so not properly calibrated, but it's not undriveable either.

As far as swaybar thicknessX1000 goes, I've had a look at other vehicles' specs and owner mods, and remain convinced it gives proper results in all cases. I set it, and build the suspension around that figure. The tough part is finding the number, as they are not usually on any spec sheet, so that leaves looking for clues on enthusiast sites of the same or similar cars, or plain old guesswork.
There's no need to issue warnings against my suggestions, if someone wants to try it they can deliberate is usefulness on their own. If I had issued warnings for every bit of pure BS a certain carwrecker was spewing, I'd still be typing them up.

[Tomte]

Sounds like a challenge.

Raised sups.friction to 300, stiffness to 12000, set diff to 90 and pulled off the swaybars. It's quick and dirty, so not properly calibrated, but it's not undriveable either.

Yes, it's a challenge and it buggers me a lot.
I will try your suggestions and see what it does. Thanks for still trying to help me. Did I mention that I had to move the CoG backwards to -0.57?

There's no need to issue warnings against my suggestions, if someone wants to try it they can deliberate is usefulness on their own. If I had issued warnings for every bit of pure BS a certain carwrecker was spewing, I'd still be typing them up.

Ok, I'm sorry. I probably tried to be too much of a smart-ass here. You're right, no need to issue a warning.

Can I raise the next issue:

Code:

jerkRPM

What does that exactly do? Apart from making the engine unwilling to rev from idle if jerkRPM is set higher than idleRPM?

[djpimley]

Yes, it's a challenge and it buggers me a lot.

Surely you mean bugs?

[Tomte]

Yes, it's a challenge and it buggers me a lot.

Surely you mean bugs?

Checking dictionary... oops, yeah, bugs me is better

[See Flat]

Deliverance...redline the sequel....

[aegidian]

Can I raise the next issue:

Code:

jerkRPM

What does that exactly do? Apart from making the engine unwilling to rev from idle if jerkRPM is set higher than idleRPM?

jerkRPM appears to set the absolute minimum RPM for the engine ie. if the clutch is engaged and the wheels are at a standstill the engine will still be turning over at jerkRPM rather than stalling.

[Tomte]

jerkRPM appears to set the absolute minimum RPM for the engine ie. if the clutch is engaged and the wheels are at a standstill the engine will still be turning over at jerkRPM rather than stalling.

Thanks for clearing that up.
I wonder what's the difference to

Code:

idleRPM

[DonaemouS]

so:

Code:

idleRPM

Is just sound pitch related? I asking cause I hit an issue with CTS-V engine hole...

[DANDY]

so:

Code:

idleRPM

Is just sound pitch related? I asking cause I hit an issue with CTS-V engine hole...

I don't believe so...
untested, but jerk should probably be less than idle, as in trying to brake to a stop without clutching in a manual car, it starts to jerk about, before finally stalling. It probably doesn't do anything with a properly set clutchRPM, which should normally be as low as possible, but not lower than idle. I had the 'hole' problem as well, I think on the XKR, which was solved by setting it higher.

[DonaemouS]

Engine inertia

Stored in engine.inertia.engine. (notice that engine.inertia.final_drive isn't used anymore) The engine takes power to spin up. For example, with the clutch fully disengaged you would otherwise spin up the engine far too fast. This slowed engine acceleration represents the engine inertia, and is specified by engine.inertia.engine (in units kg*m^2). A typical value may be 0.09, but I'm not too sure about this.

This is how engine inertia work in Racer. Is the kg*m^2 the same unit used in Redline?

[Sponsored content]