Like your FFB strong? Think again..

[Niels_at_home]

Heyas,

I've been around this simracing scene now forever and it seems the vast majority of folks like to run very high force feedback settings. Then you often read claims how great it feels and how they can even feel the effect of flies building up on the windscreen through these FFB settings (sarcasm = on). Some even feel when the coffee is done in the kitchen (sarcasm = still on)..

Force Feedback talk
Good sims will use complex physics engines to calculate the force feedback. Part of the feel is the tires, part is the suspension geometry, and this all changes with steering angle and things like how much weight (load) is on the tires.

If you'd only feel the tire part of the force feedback, the forces would be near ZERO by the time you've applied the optimum steering angle to get maximum grip. Thats just how tire physics (pneumatic trail) works, give or take some margins. Counter intuitive eh, but its often near true!

Suspension geometry usually adds a chunk of force, but still it is higly likely that the force you feel in a racing car drops off after a certain amount of steering and by the time you've applied the optimum steering angle for max grip, force through the steering wheel might be about half of the maximum.

That means when the steering goes light in a racing car, you might not be understeering at all, you might be right at the maximum grip levels!

Problems with sims
Even when your favourite sim has good tire and suspension physics, that doesn't mean your wheel (G25, Fanatec, Thrustmaster etc) will behave properly. The physics engine will just calculate a torque or a steering rack force, perhaps during a lap this value is between 0 and 5000 Newtons.

At some point the sim will have to tell your FFB steering wheel what force it has to apply. You can imagine there is some scaling required. Ideally the 0 .. 5000Newton that occur on the track should be the range where your FFB wheel responds between 0% and 100% of its capable force.

This scaling is NOT DONE AUTOMATICALLY by iracing / LFS / rFactor&Realfeel !! You have to scale this yourself. In iRacing and LFS its the force feedback strength in the menu, with rFactor&Realfeel there is the realfeelplugin.ini 'Max Force' entry to adjust.

When you scale the forces down too much, you get weak force feedback. When you don't scale them down enough, you're asking more from your wheel than it can do, so at some point the forces will be trimmed. This is called 'clipping'. If your wheel has a maximum of 100, you can't make it do 200. So anything you ask from it above 100, you won't feel a difference.

First chart: Scaling the force feedback

The blue line is how a good simulator might calculate the force feedback strength versus your steering wheel angle. It is what it *wants* your FFB wheel to do.

Lets say proper scaling means you have to run the ingame FFB strength at 20% which is not uncommon at all. In this example, 20% strength matches the maximum occuring force feedback of the physics engine to the maximum feedback your wheel can put out.

If you'd use 100% ingame FFB strength, you would only have a small part of the desired curve fit below that line. That is bad.


Second chart: Results

This chart is the results; what YOU feel on your wheel. You can see that running 100% FFB strength tries to make your wheel 5x stronger than it really is. It just can't do it, so you feel the red line. Apply steering, forces max out *very* quickly, and as you steer more, the force stays maxed out. It hasn't got the power to follow the desired dotted line. This is what many simracers do probably without knowing it. The force feedback isn't detailed at all, its almost on/off.

The green line is what you get with proper scaling. It follows the desired shape exactly, but now with the maximum occuring force nicely matched to your wheels maximum deliverable force.

Now why does 100% still feel stronger than 20%, despite the maximum force being the same? At 100% you have to grip the wheel like mad because any slight bump or minimal steering will result in maximum force. Once you're steering, the forces are mostly constant, but going near straight you feel strong jolts left/right as micro bumps / steering inputs cause these forces. Wheels setup like this are often unstable and will oscillate when you let go of the wheel on the straight. For some reason lots of folks think this feels good.

So how do I set this up correctly then?
Thats the hard part. Even on the same car if you change the car setup (caster, downforce) the numbers will change. On a flat street circuit the values will be different than in a fast banked oval.

Realfeel with rFactor is the easiest, you can drive the car on a track, look at the telemetry, and the 'Steering Arm Force' that you see in a hard corner is your desired Realfeel.ini "MaxForce" entry. Don't look at short spikes of force, go for a sort of average force you see in a fast bend.

With other sims its a matter of getting a feel for what your wheel can do. Once you know what its maximum strength feels like, lower the ingame FFB strength until you clearly are below this maximum force. Then add force slowly until you get to the point where it seems saturation occurs. I can't help much, the FFB strength will typically be between 1% and 100% :S. Try the maximum force in a steady corner, because on a straight, the sudden left/right jolts of too strong force feedback will make it feel subjectively stronger than it is.

Conclusions?
Current wheels like the G27 are fairly weak; they can't really deliver a strong force. It is very easy to try and ask too much from the wheel. It might also lead to overheating and faster wear!

I've tried to show you what force feedback scaling does and how running too strong force feedback settings can numb down the detail you feel, even when it *seems* like the effects are nice and strong.

Everybody is free of course to use whatever setting they like, but at least this shows that people running very high FFB strengths claiming they get lots of detail and a good feel for the car, are wrong from an objective point of view.

 

[Dariusz Swiderski]

Brilliant explanation Niel, it gives me something extra to think and try again. In my 2 1\2 years of sim-racing i was never fully satisfied how my G25 feels and have been searching ever since for that ''perfect'' feel. Started off like most people do, everything maxed out thinking stronger is better. But i didn't like it at all and started my quest for a better feel which basically means going weaker and weaker. Now i,m at a point which i though feels pretty good but your article gives me a lot more clear picture how the FFB works so i,m off to do more testing.

:thanks: very much

 

[Valerio Vinassa]

THANKS Niels nice advices

 

[Alexander Rhodes]

Another point on the same subject, the sims and games are also giving incorrect data, as well as allowing you to setup the FFB too strongly.
For example;
The reason a car would pull you onto the grass, when you get half a tyre onto it, is because the wheel loses traction, causing a toe'd wheel to turn the car. Or if the camber is high, the grass will allow your wheel to sink, causing the opposite rim to hit the tarmac at it's closest point, releasing it's traction for long enough to allow the sinking wheel to pull you in.

Or if you have a stiff or locked diff, and your in a front/4 wheel drive car, you turn in off the power, you apply the power before you straighten up, and no no no, the front wheels will only want to go straight towards the direction of G, pulling the wheel into the direction of G.

Or on curbs, with camber and toe, and stiff suspension (not common on soft setups!) The wheel on the curbs will hop from 1 lip to the next, and every time it does, the wheel wants to distribute the force along it's walls, and phsyics intervenes with the opposite effect being the force of landing on the lip will push the wheel one way or the other.

But we don't see this quality of physics on normal "sim" games, nor do we see it to great extent on any of the best sim's. The reason behind this is not only tyre modelling data, but surface, geometrics and material simulation. Once we can ace all three, and get it to work consistently with suspension point of axis and weather data, then we can worry about realistic FFB. For now, don't expect anything more than a workout.

FFB was a gimmick when it came out, much like "dual shock" "rumble packs" etc... it's designed to make sales, and therefore games want to incorporate it, and then people creating simulations, want to test it, try to improve it, and it's all still too soon. If you want to feel what going around a track feels like when your on the limit, go do it. You'll be shocked at how different any FFB setting will feel.

So if you want to work out, use it, like I do If you want real feel, go on track, you'll not find it in any of todays sims.

Note: I am not saying the wheels will not give accurate FFB, or that they are too weak (for a lot of cars and setups, wheels like Frex and LB's new one will be more than ample enough). What I am saying, is that sims and games cannot give out the correct data to your fingertips (and on the limit, it is fingertip feel, not bicep feel).

P.S. I'm a 24 hour specialist karter, I do stints of 2~3 hours at a time, and only once have I ever found the steering heavy (bent tie rod ) I have also tested Ginetta's and Porche's and can say that they will never tug the wheel, never wobble the wheel and never ever snatch the wheel out of your hands, the tyres will lose grip first (due to the steering ratio and power assist in all modern race cars post 1990's).

Summary: It is not the wheel's output capabilities at fault, the working range (below ceiling) should be more than enough for going around a track quickly. It won't however be strong enough for impacts until certain health and safety laws have been forgotten about .

If anything is to fault, it is the simulation, for giving inaccurate motions to the wheel's motors in an attempt to simulate what the whole car is doing, as opposed to the front wheels. So no the G27 is not at fault, todays racing cars will not feel as strong as that can get (excluding impacts)

Well done to Neil and Brian for pointing this out to everyone, I thought it was common knowledge

 

[Xosé Estrada]

Thanks!

 

[Gary Corby]

This is brilliant stuff. Thanks so much. I've a question for you:In a game, I can set the FFB down to, say, 30%, and leave the wheel driver at 100%. Or I can leave the game at 100% FFB and set the wheel to 30% force. The results are totally different. From playing around, I deduce wheel drivers do their force reduction *after* clipping, because with the second method it feels to me like a clipped response reduced to 30%. Do I have that right? If I do it the first way though, with the game at 30% and the wheel at 100%, then the wheel is very loose on a straight and tight on a turn, from which I deduce I'm seeing the full dynamic range. If I have this right, then why don't wheel drivers have their own auto-scaling feature rather than clip? It would be so easy to code, and could "error correct" for clipping. For reference, I tried this with F1 2010 and a Logitech DFGT.

 

[Senad Subasic]

I think that's about right.

As far as auto-scaling goes, I think it's better if it's not there. This way you can pretty easily get the right settings, but still have the option to get clipping, for whatever reason you might have.

 

[Timo Birnschein]

Nice article! I'll definately try this since I was thinking about a weaker setting for some time now. During the last couple of races I was constantly turning down the strength of the G25 wheel to reduce the amount of stress for myself during that race. Helped! But finding the perfect setting for the wheel sounds like a nice challenge

 

[Gary Corby]

Thanks for the answer. I'm pleased I managed to understand that much! On the auto-scaling, it seems to me this is analogous to producing music, where with production software you can define a frequency envelope within which the signal is scaled. Applying that idea to the driving wheel, if you want the raw linear signal as a default, fine, but as an advanced option it seems to me some sort of logarithmic envelope would give the driver a tighter, more comfortable wheel for zero newton straights, while still producing the much tighter feel at 5K newtons, without ever resorting to clipping.

 

[Sérgio Marques]

Just two words and your FFB problems are solved: NetKar Pro!

 

[John Cunningham]

yeah but... its the only exercise i get.. (sarcasm=still on:tongue

 

[Alexander Rhodes]

NetKar pro is getting closer, but it's still a long way off. It will be a few years for sure before we can say we're getting close to real world simulation.

 

[Niels_at_home]

I haven't tried it for a long time. Lets not discuss sim realism; thats another big can of worms to open! What might be relevant is that NKpro has a built in 'ffb clipping' check which would be very usefull to have in other sims! Kudos to Kunos on that one..

 

[Alexander Rhodes]

Yeah, the clipping check should be a feature of all games now for sure

Another question on the clipping, does anyone know if all wheels start to degrade at the same point? It would be interesting to know if 1 wheel would could avoid clipping at a higher strength than another. Or if it would be possible to set resistance strength and FFB strength separately. Now that would be nice

 

[Niels_at_home]

You are potentially opening another can of worms Alexander

Sim software using force feedback wheels, uses standardized communications, I believe the force output is between -10000 and +10000. How much you get depends on the sim and FFB settings, so with proper scaling, a high speed corner would approach this 10000 limit ideally.

How wheels respond in force within this range could depend. I've measured the G25 to respond almost linearly between 0 and 10000, using 100% Logi Profiler strength. So as soon as you use more than 100% in the Logitech software, you add clipping even if your sim sends a nicely scaled non clipped value to the wheel.

Wheels aren't necessarily linear in response and some might clip at 100%, some sooner, some later, depends how well they've done their homework

 

[Alexander Rhodes]

Yeah, the software is easy enough to measure, and I'm not sure if any other brand has a profiler, such as logitechs. But is the output entirely dependent on the sim? I've seen a few motors in my time, and know that every different model has differing amounts of torque and power, and that means that out there are going to be some variations. But how do we know if the motor is at max torque without knowing if it's at the maximum draw? And how do we know it's using it's max output if we're totally depending on % outputs from sims? The clipping is entirely software/driver/firmware dependent, so surely that means the software could be curbing the maximum output the motor has to offer right?

I know that testing this means writing a new driver and program and maybe firmware to manage the wheel, but even though it would be tough, it would be interesting to see if the brands have left anything in reserve .

 

[Marco Boemi]

Thanks for sharing your knowledge Niels. Do you think the "Curva grande" at Monza is a good place to find the maximum force a wheel can do? My arms say yes

 

[Dave Stephenson]

yeah but... its the only exercise i get.. (sarcasm=still on:tongue

what he said... (sarcasm=not on :eek

 

[Ryan Callan]

Great article Niels, always enjoy your technical insight.

 

[George Pilkington]

Great Thread Guys, thanks for all who have written so imformativly.
I'm going to reduce the profiler to 100% on my G27.
I did not race at indy 500 (RD one) this year as the car was terrible to hold strait on the straits... now I understand why! I'll do some experimenting too. : )
Regards
George

 

[Abdul Al-Amry]

What's Clipping?