Reality vs. FFB vs. Assetto Corsa

Hi guys,

I've been thinking about this matter for some time and I'm really interested in your opinion.
The thing which inspired me to write this thread was some angry voices after AC TP release that Elise SC "has got too much understeer, too much grip and it doesn't drift" (despite the fact that real car doesn't either...) and also some opinions that FFB isn't strong/good enough.

What am I thinking about? Well, I'm interested in your feelings about force feedback in game vs. reality.
Let's suppose that we've got 3 cars in Assetto Corsa (or actually any other sim):
- first car with no power steering
- second one with hydraulic power steering
- third one with electric power steering

We suppose that their real counterparts has got following traits: 1st one has got the best 'road feel', 2nd one good 'roal feel' and 3rd one the worst 'road feel'.

My question is - should it be presented in the game? Should the FFB of the first car (without PS) be crisp and sharp with more power on the wheel and the FFB of the third car (with electric PS) should be numb and should NOT give you that many informations about what's going on with the car's wheels?
In example the difference in 'road feel' between Porsche 911 997 with hydraulic PS vs. 991 with electric PS which all car magazines wrote about.

I feel it could be problematic as we saw with TP because most of the simracers are used to racing cars and when FFB of street Lotus Elise didn't match the FFB of GT1 or F1 car they were used to they thought it was just 'not that good'.

What do you think about this problem guys? How Kunos (or any other publisher) should solve it?

Same question is with the sounds - if Elise SC sounds poorly in the reality should it sound ingame like Ferrari's V12 or Toyota 1.8L engine? (because I also saw people unhappy that Elise's engine didn't sound like racing car's engine)
 
What am I thinking about? Well, I'm interested in your feelings about force feedback in game vs. reality.
Let's suppose that we've got 3 cars in Assetto Corsa (or actually any other sim):
- first car with no power steering
- second one with hydraulic power steering
- third one with electric power steering

I agree this should be considered, and can be achieved somewhat.

Once tyre forces leave the tyre contact patch, and get adjusted due to the CP vs suspension kinematics, they end up on the rack.
How they then get from rack to wheel is another system entirely and ideally it could be simulated with a bunch of coefficients you could tune.

Usually steering ratio is in there already, steering wheel diameter can be measured so the force at the tip of each cars wheel could be calculated in theory.

You could then add layers of damping, inertia and assistance depending on characteristics of the car. Ie, is it a lightweight Elise or a heavy Rolls Royce. Power assistance or not etc.


It's something I've thought about a lot and is missing in most games. Many avoid the issue by offering only one certain type of cars so it's hard to gauge the variables. Since AC is offering road cars to race cars it will be something they should probably look at.

I'm not sure if it's that important because in GT5 for example I don't really care too much that they all feel about the same, the FF gives enough info in any case to drive better.

Should heavy cars 'feel' different to light ones? Well I guess so and in theory it's not hard to achieve that... it just needs the authors to balance the content appropriately.


Ultimately we hit another barrier though which is FF wheels can only provide so much force. A road car with many turns lock vs a race car with fast lock and much higher forces can't be reproduced exactly so do you scale road car forces down so race cars feel appropriately heavy? Or do you let road cars feel right but then race cars can't feel much heavier?

Ie, a kinda auto force gain control hehe.


Needless to say it's another minefield but one that someone should approach at some point because imo it'd make a great difference to a game like this to have an Elise feel genuinely feel-some against heavy road cars like the 1M BMW.



Many people shut off the FF effects in the Logitech profiler but really they should be left at 100%. They only work if the content developer sends signals to those channels to do something and it is in the controller that those high frequency effects are generated and give great feel.
Turning them off is probably losing you programmed in weighting, damping and feel that perhaps AC is requiring you have turned on.
The fundamental forces from the Fy of the tyre about the steering axis into the FF are only part of the FF feel. They are also relatively low frequency and can't be updated fast enough to give a great feel. There is a good chance that KS have programmed the other subtle damping and weighting effects via the DirectPlay effects in the wheels controller, so those effects should be on really.

I'm not sure if AC can force those effects back on if they are turned off but it'd be important to have KS clarify on those things. In theory even if they are turned on, a good game should set them to zero if they are not required any way.

Dave
 
I agree this should be considered, and can be achieved somewhat.

Once tyre forces leave the tyre contact patch, and get adjusted due to the CP vs suspension kinematics, they end up on the rack.
How they then get from rack to wheel is another system entirely and ideally it could be simulated with a bunch of coefficients you could tune.

Usually steering ratio is in there already, steering wheel diameter can be measured so the force at the tip of each cars wheel could be calculated in theory.

You could then add layers of damping, inertia and assistance depending on characteristics of the car. Ie, is it a lightweight Elise or a heavy Rolls Royce. Power assistance or not etc.


It's something I've thought about a lot and is missing in most games. Many avoid the issue by offering only one certain type of cars so it's hard to gauge the variables. Since AC is offering road cars to race cars it will be something they should probably look at.

I'm not sure if it's that important because in GT5 for example I don't really care too much that they all feel about the same, the FF gives enough info in any case to drive better.

Should heavy cars 'feel' different to light ones? Well I guess so and in theory it's not hard to achieve that... it just needs the authors to balance the content appropriately.


Ultimately we hit another barrier though which is FF wheels can only provide so much force. A road car with many turns lock vs a race car with fast lock and much higher forces can't be reproduced exactly so do you scale road car forces down so race cars feel appropriately heavy? Or do you let road cars feel right but then race cars can't feel much heavier?

Ie, a kinda auto force gain control hehe.


Needless to say it's another minefield but one that someone should approach at some point because imo it'd make a great difference to a game like this to have an Elise feel genuinely feel-some against heavy road cars like the 1M BMW.



Many people shut off the FF effects in the Logitech profiler but really they should be left at 100%. They only work if the content developer sends signals to those channels to do something and it is in the controller that those high frequency effects are generated and give great feel.
Turning them off is probably losing you programmed in weighting, damping and feel that perhaps AC is requiring you have turned on.
The fundamental forces from the Fy of the tyre about the steering axis into the FF are only part of the FF feel. They are also relatively low frequency and can't be updated fast enough to give a great feel. There is a good chance that KS have programmed the other subtle damping and weighting effects via the DirectPlay effects in the wheels controller, so those effects should be on really.

I'm not sure if AC can force those effects back on if they are turned off but it'd be important to have KS clarify on those things. In theory even if they are turned on, a good game should set them to zero if they are not required any way.

Dave

Obviusly modern wheels (those for mass market I mean) have their limitations.
I wonder is that possible to have different feeling depending on car and it's PS or lack of it (and additionaly on type of drive - RWD vs. FWD - torque steering like in focus rs) WHILE still giving the gamer info about what's going on with the car - because as earlier was said FFB have to substitute somehow i.e. G-forces...
The main thing is to feel the difference between no PS/PS just to give you more immersion.
The question is - are the modern FFB systems able to give you so many informations at the same time?

Btw. this auto force gain control was also something I thought about some time ago... :D
 
I think PS is something we can simulate ok.

I've tried programming PS for the Racer Free Car Simulator using numerous methods but I always seem to get something that doesn't feel right.

We have to consider that a steering wheel on a real car turns in response to the Mz force turning the wheels to point where the Mz returns to near zero (equilibrium)
That force can be very high, but we don't feel it being that high due to PS.

But how do we get the self-correction of high Mz and rack forces, but also get the light feeling of PS when we want to turn.

It's kinda a catch 22 to me. I need to think about it more but PS comes into effect when we resist the Mz, but how do we measure Mz resistance?

Maybe we need to monitor wheel turning speed and Mz, if wheel speed is low and Mz is high then that means it's being resisted? At that point we can turn Mz down to simulate assistance?

Then we release the assistance under certain conditions, like if Mz returns to near 0 then start turning back up the Mz force to the original unassisted level?

I guess we then get into feedback loops of equilibrium where there is a sweet spot.


What would probably give us good clues is looking at the valving in a power assisted hydraulic rack and look at the logic of it's operation. I guess it's just ramp rates based on occlusion of a hole as the rack moves. Not sure how they sense when you resist Mz though.

Hmmmm...

And the logic of the electronic systems, I guess to find out we can decompile the ECU data to take a look at those?

Dave
 
What would probably give us good clues is looking at the valving in a power assisted hydraulic rack and look at the logic of it's operation. I guess it's just ramp rates based on occlusion of a hole as the rack moves. Not sure how they sense when you resist Mz though.

I was curious so I looked up a fairly generic explanation. There's a torsion bar between the wheel and the rack - essentially tuned so that it twists a certain distance based on how much force the driver applies to the wheel (assume the rack is fixed for now). At neutral position, the hydraulic valves are equal in both directions - as you twist it they bias to one side, and the power steering applies torque to the rack to try to line it up with the wheel. It's essentially a force multiplier - so if the wheel Mz is pushing one way, you push back with 1/n of the force, and the power steering provides the other (n-1)/n. Once the torsion bar is twisted as far as it goes, it hits a bumpstop and the power steering applies max force and you can add to it with your arm strength.

It seems like the easiest way to implement this would just be to have a curve LUT, with Mz on one axis and wheel force on the other - probably near dead center it would be close to 1:1, and then ramp downward so that for a little bit of wheel-force you counteract a lot of Mz. I haven't had the opportunity to abuse power steering so I don't know how it ramps up to allow for when you hit an obstacle (eg. if your tire runs into a curb, no amount of force is gonna turn it farther).
mYFICpb.png

How 'light' the steering is then depends how much the PS is reducing the needed force. You need a little bit of feedback of course to be able to feel what's going on.

This also hints at why it's inferior to a direct connection; the torsion bar means there's an additional spring constant between you and moving the tires, which damps the steering feel slightly.
 
Haha, a LUT. You know in Racer we'll need to be doing a polynomial for that ;) :D

But for a general approach then that is an interesting arrangement. You get near un-assisted tyre forces around low-load and dead ahead but then as the forces build up *and you resist them* then you get assistance.

But how do you emulate the ability for the mechanical system to detect resistance in what the rack wants to do and what you as the user of the steering wheel want to do?

The FFB wheel is linked directly to the rack so no way to sense that differential. You'd have to calculate it from wheel position and velocity and rack force perhaps? Detect resistance under a rising force and a non-rising wheel position, or something along those lines...


Hmmmm. I wonder if this is something you could elegantly apply as a post effect then, or if it really needs to exist in the entire physics/FF system so it can apply the torsion effect and inherent lag into the FF system.

Hmmmm

Dave
 
But how do you emulate the ability for the mechanical system to detect resistance in what the rack wants to do and what you as the user of the steering wheel want to do?

Hmm on reflection I'm not sure it has to. It doesn't matter which end of the power steering is actually applying a torque, it resists it according to where the wheel is. So when you drift a bit and the wheel suddenly wants to spin the other way, it can only apply a smaller force to the wheel side of things - probably still ample to spin it quickly, but in principle, if the power steering limited force enough, then when it tried to jerk the wheel around, it could only apply a small force to the wheel before the inertia of the wheel made power steering push back. It's stable either because the aligning moment is 0, or the steering wheel's force is proportional to the aligning moment (not equal because of the various mechanical advantages, plus power steering itself)


I'm definitely curious how much actual play there is in this torsion bar. Is it a couple degrees of flex during normal driving or what? If it's small it can probably just be written off as less significant than the lag in the FFB system. Maybe you can feel it with the car turned off, with the wheel having a bit of spring in it before the tires move? Ripping open the power steering and looking at it would be the only way to be sure.



For electric power steering, I'm not sure you can generalize. In drive by wire I'd expect a low-pass filter to be adequate, since it's only gonna be sampling the force at some rate. If you have a couple low-voltage motors (like LEGO ones) you can put them into a circuit and input on one motor translates to the other this way - it's only responsive to low frequency changes. I don't know how it measures the power assistance need if it's an electric motor hooked up to a mechanical steering system. Maybe you get the opposite - low frequencies are muffled and high-frequency info passes through untouched. Aside from complaints by porsche enthusiasts that seems ideal - it'd help you with the forces of low-speed steering, but still be sensitive to sudden changes.
 
I'm just not sure how it looks from a code perspective hehe.

My Z4 has electric power steering and via diagnostics you can log the steering torque. I might have to do that one day and see what kinda values it's throwing out vs FF ones.

It does make you wonder if the KS team have even looked at this yet though. We've all sampled the Elise but fortunately it's a very basic car with no ABS, TCS, ESP, PS or assisted brakes even. Very easy to simulate easily for a tech demo I suppose.

A 1M with turbo and M-differential and PS and ABS, ESP, TCS etc is probably a whole different kettle of fish with many layers of extra coding to simulate those elements.


I doubt there can be much play in the steering rack. I think the tolerances are super low. Most cars have a non constant velocity universal joint on the steering shaft to rack too.

Whatever effect is occurring obviously does so on a very sensitive basis. Perhaps very old systems are quite a bit more laggy though.


Assuming you could program this then I assume you'd just monitor steer velocity and the aligning torque, and use a raft of coefficients and lag values to tune the way the systems cuts aligning torque based on a resistive force detected.

Hmmmm

Dave
 
Ah ok maybe the current ones do.

Would be interesting to see the ABS code since we get loads of different types on modern cars.

Wonder if it's using wheel speeds only and is impacted by driving up/down hills :D

Dave
 
My Z4 has electric power steering and via diagnostics you can log the steering torque. I might have to do that one day and see what kinda values it's throwing out vs FF ones.

that would be interesting but the torque sensor usually measures only up to 8Nm or 12Nm, so that is fairly limited... the torsion bar is limited to a specific value with "bumpstops" so after that when you hit those it just means "full power" for the steering assist... So my guess is you wouldn't be able to actually get all the infos ("extreme" situations) you'd hope for when monitoring the steering torque from your car...
 
Hi guys,

I've been thinking about this matter for some time and I'm really interested in your opinion.
The thing which inspired me to write this thread was some angry voices after AC TP release that Elise SC "has got too much understeer, too much grip and it doesn't drift" (despite the fact that real car doesn't either...) and also some opinions that FFB isn't strong/good enough.

It does drift. But you have to know how. You must force its back out but then it works perfectly well. I have often driven the real Elise with 156 PS and AC have done a very good job.
 
It does drift. But you have to know how. You must force its back out but then it works perfectly well. I have often driven the real Elise with 156 PS and AC have done a very good job.

Yeah, I know it can drift (helping it a little with brakes also does the job) but that's not the case.
There was a group of people who thought that if it's RWD it have to drift. So they wrote that it's unreal that RWD car grips so well and doesn't drift when even when pushed.
And the best answer was showing them the same real Elise SC in Top Gear in which Clarkson couldn't drift as well (until the guys from Lotus showed him how to)...
I was just compering the situations with this above as an example. That was the case ;-)
 

Latest News

Online or Offline racing?

  • 100% online racing

    Votes: 87 7.4%
  • 75% online 25% offline

    Votes: 123 10.5%
  • 50% online 50% offline

    Votes: 170 14.5%
  • 25% online 75% offline

    Votes: 327 27.9%
  • 100% offline racing

    Votes: 459 39.2%
  • Something else, explain in comment

    Votes: 4 0.3%
Back
Top