Calibrating your force feedback wheel using wheelcheck and ffbclip

Force Feedback steeringwheel calibration​

You should always calibrate the wheel and forcefeedback multiplier in the game to ensure you get the best, unclipped forcefeedback you can achieve with your hardware. The software mentioned in this thread is attached at the bottom.

1a. Find the most linear setting for your hardware
There is some discussion about what is the best method to determine this setting. For most users, leaving it at the default value seems to be close to optimal. The issue is touched upon in this post.
1b. Using a Look Up Table (LUT)
1c. Find your minimum force setting
2. Find the maximal clipping free force feedback multiplier in Assetto Corsa using FFBClip app
3. Putting it all together

1a. Find the most linear gain setting for your hardware
There are several methods to determine this. One is the "pole and scale" method, where you mount a pole on your wheel there one end rests on a standard kitchen scale. One great advantage with this method is that you also measure the max torque of your wheel. It is also likely to give you the most linear response, since it is a static test.

Using Wheelcheck:
Use the "log 2" setting, let the program run through the tests and create the log file. It will be placed in your "documents" folder and be named <Log2 "Timestamp".CSV>. Import this file to a spreadsheet diagram using Force as Y axis and DeltaXdeg as X-axis. It is useful to scale the DeltaXdeg value from 0-100, meaning the maximum DeltaXdeg value in the CSV file counts as 100 and scale the graph accordingly.
Do this for several settings in your steeringwheel control panel, especially gain and any settings that can affect linearity. Consult the manual or ask in the respective thread for your hardware how to do this if you are unsure.
For most modern wheels the step tme value needs to be lowered to around 100ms in Wheelcheck to provide the most accurate results.
Here's where you find the Log2 test in WheelCheck:

When you find the setting with the best linearity, note the maximum corresponding force equivalent (the scaled x-axis). Check the example below:
If you are unsure how to get the log file into a pretty graph there's a very nice post by @mms here:
Getting the Wheelcheck log into Excel

1b. Using a Look Up table
This topic has it's own tool and is discussed at length in this thread

1c. Finding you minimum force setting
Note! This is only applicable if you are not using a Look Up Table as discussed in 1b!
Now to another setting that is important for Force Feedback linearity, the "minimum Force" setting. You find this setting in Assetto Corsa -> MAIN MENU -> OPTIONS -> CONTROLS -> ADVANCED -> Minimum force. This setting is nescessary because most wheels have a "deadzone" around zero force. That means that small forces from bumps going down a straight might be lost since your wheel can't reproduce these small forces. The "Minimum Force" setting "lift" these small forces to a level where your wheel can reproduse them. So for instance; going down the Monza straight you get some small bumps that are around 2% of FFB. Your wheel doesn't start to reproduce forces until 10% (it has a 10% deadzone) so these forces are lost. Now, if you set the minimum force setting to 10%, those small forces are lifted and start at 10%, so the small bumps are now at 12% and your wheel can reproduce them! Voila, your wheel no longer have a deadzone!
The most accurate way to determine the minimum force needed is to take a car with a lot of feedback around center, go on a bumpy track and check that you feel a "buzz" or some road texture on the straights. I usually take the KTM X-Bow R on Magione. The back straight has some nice small bumps that you should feel in your hands.

2. Using FFBClip app
Install the app, load up your favourite car and track combo. It's that easy! The app will go to auto mode for any new car/track combination you use to determine the correct gain for that combination. It will load up the appropriate gain value whenever you go back to a track/car combo that you have used before. For more info on the FFBClip app, visit the FFBClip thread

Now, you might wonder why we "allow" any clipping at all. The reason is the filter built into the ffbClip app. That will ensure that you still feel most of the bumps even if you are very close to clipping, thus maximizing the available force feedback strength.
And why do we need to maximize the force available? Because most consumer force feedback wheels on the market are nowhere close to being able to reproduce the realistic forces required, so we might as well get as close as we can. Now, some high-end wheels are strong enough, and for those few wheels (Leo Bodnar wheels and the like) there is a reason to limit the maximal hardware force generated.

So, an example to show why mild clipping is a necessary compromise for consumer wheels:
If we look at the example here we have a situation where a car's FFB multiplier is tuned "perfectly" in a classical sense. The graph stays below clipping at al times. The yellow graph represent the filtered force feedback, and give an indication where the FFB multiplier for this car would end up, somewhere around 80%.


In the next example we have raised the multiplier significantly, and the cipping here is very severe. Remember that anything over the red line in the graph is clipped, so in this example we will feel no bumps or detail in the wheel at al, just a constant force pulling the wheel to one side. If the car start to slide out or we hit a bump in this situation we will not feel it in the wheel at all.


In the last example we have tuned the ffb with FFBCip, and as we can see the yellow graph is very close to cipping, and some of the detail is lost! But, this is actualy desirable behaviour for most wheels. If we take a closer look at what's happening with the wheel here versus the first example with no clipping: We are turning hard, and hit bumps or a kerb that vibrate the wheel. This is what we want to happen, right? So, in this example we are turning hard, there is even more force here than in the first example, and the parts of the graph that goes below the red line means that we still get the vibrations in the wheel, it's just the top half of the bumps that are chopped off. So, we have increased the maximum force available AND we still get the buzz in the wheel over bumps and kerbs! Double whammy bonus! The increased multiplier means that small forces are amplified as well, such as small undulations and bumps when going straight, and minute forces of wheels slipping and effects of setup changes.



3. Putting it all together.

PLEASE remember that turning down the gain due to clipping doesn't mean you lose any force, you are just bringing the force to a level where it contains information. If your wheel is clipping it doesn't matter if the gain is set to 150% or 45%, there won't be any more force coming out ot it in either case. So, if your wheel can output 4Nm of force, with your level set to clipping it will output 4Nm more of the time but it will be constant force with no information in it. With the level set correctly you still get 4Nm but all the information is there.
Let's use the T500RS in the first example, and assume I chose the 100% gain setting in the driver control panel. This means my max hardware force is reached at 55% FFB signal from the game. For a certain car/track combination FFBClip came up with 107% as the optimal gain setting. So, I now have the wheel gain and game gain working together to bring me the maximum force with just the right amount of clipping for maximum enjoyment and realism!

 

This needs to be stickied!

 

Great post

Is it safe to assume that the sim is sending a linear signal to our wheels? Some other sims such as Iracing didn't for years and only more recently have enabled the linear option.

 

Sure feels linear. I think the philosophy of Kunos is to have the wheels do the signal processing.

 

Great post!

I'm pretty happy with my current FFB settings but I'll give it a try when I have enough time.

 

I did the Wheelcheck on my G27 and got some interesting results.

100% Spring, 100% Dampening


It's more a sinoid curve than linear. Although up to 50% of normalized measurement, it is pretty linear:



When doing the same measurements with 0% Spring and 0% Dampening, there wasn't a lot of difference. Notice how with both settings there is a sudden bump in the high/extreme values. This is around 0.8 in normalized values, around 135 in real values.



For this last one, I added two linear regressions per data series: one from value>0 & value<0.5, and the second value>0.5 & value<0.9: it seems like there is a real "kink" in the measurements.



Any suggestions what a G27 driver should use as the base FFB percentage? I've always used 100%, but maybe 80% is better?

 

I would say the 80 is better, even though you will lose that little extra bump up top. Try them and see!

Very interesting measurements, though. I wonder if all G27 are like that?

Edit: Quick google-fu show that that is indeed typical for the G27.

 

@mangal great job with G27.
From your diagrams we can see how bad this wheel is.
If you want to avoid the "dead zone" around zero , then you have to choose 110%. But then you will encounter the strange behaviour at high forces.
If you want to avoid that completely then you choose 80%. But then you have a huge hole around zero.

For me the "dead zone" is totally unacceptable so i choose 110% and all spring/dampening at 0%. (actually i have it at 107% as many has mentioned that here at forum). Maybe you can also add this measurement.

 

I tried with my T500RS, since a gain of 60 is by far the most linear, the 100% gain showed less deadzone. It was very hard to detect for the T500, but might be easier with the larger deadzone of the G27.
I wish we had a "minimum force" setting like iRacing. These measurements provide ample evidence of why we need such a setting. We are mere mortals, and out wheels can't keep up with the demand from high-end simulators.

 

Here you go. I added in the 107%, as well as the 101%, as that is also one of those "anti-dead-zone" mythical values:


The 107% is pretty much the same as the 110% once over 50% of normalized measurement, and the same with 100% and 101%.

 

If it's anything like the DFGT, once you go beyond the top end "step" in the FFB power the wheel feels like a cog. my hunch is this is the electric motors themselves and not by design.

I'd probably go with 110 or 107 in the profiler, and set the gain in the sim so you're never sending the wheel more than an 80% signal, thus avoiding the "step".

For me a significant deadzone is a killer and should be avoided as you won't feel the rear of the car move around until it's too late.

 

Hopefully Kunos will add something like the Steering Torque Minimum(rfactor2) or the minforce (Iracing)
in the future so the 107% semi fix is not needed.

I bet they will.

 

I set my G27 Profiler to 107%, then set the FFB gain in AC to 85% (or rather, left it where it was). I tested in the Lotus Evora GTC at Mugello, and got to a FFB recommended of 62%. I tested that, too, and that suggested 63%, so I gave it one tick more.

Then tried Donington, and while a quick check suggested 67% there, I kept going, and when I looked up I had taken 2 seconds of my PB.

All in all the FFB is a bit lighter, now (before I could lean on the wheel a bit more through fast corners), but without question it is an improvement in the sense of resolution and variation. I'll have to get used to the lighter feel, but overall it gives a much better sense of what is going on.

Thanks a lot, Atle. This has been very enlightening.

 

Thank you for the encouraging words! It's good to see that most people get a better experience. Now, let's hope the devs watch these threads and give us the minimum force slider sooner rather than later!

 

@mangal , as we can see from results from the G27 (thanks for that) , we can see that as you increase the FFB percentage from driver , the "dead zone" is getting smaller and smaller.
So i was wondering , is there a value that will give zero dead zone (forces from almost 0 degrees of rotation) . Maybe a 120 or 130%.

And @Atle Dreier , i don't know if what i will say is possible , but i think the perfect app will be the one that reads clipping and automatically alter FFB gain in order to avoid that or allows a value of let's say 5% ...
Nevertheless , i truly thank you for your awesome app and also awesome guidance .

 

Liakjim: Sadly it's not possible to write to those values, I can only read the Main menu multiplier. I don't even have access to the multiplier in car setup, so I have to rely on users having that set to 100% as per the instructions.

 

Ok, here's my DFGT, proving its superiority over the G series!

 

@djkostas i thought you were playing with a G25...
Truly excellent results, makes me mad about my G27 performance around 0.
I would choose 100% as most linear.

 

@Atle Dreier, thanks for this! Quick question, from the example of @mangal, would your app's recommended multiplier still be accurate since he set his Profiler value to 107, and not 100 as your app requires? My graph is very similar. and I believe most G27's have similar.

 

It would be nice if you could somehow implement/input your personal wheels best linear setting direct in the ffbClip app so that you get the correct setting without the need to use the calculator.
So for example for my DFGT i use a 0.9 factor to get to the best results, i put that into ffbClip and then magic happens and the result is the correct FFB setting you can use for the car!
This is just wishfull thinking and i do not mind the extra step manually.

Thank you for your work and explanation, it has been truelly helpfull getting the best feeling in my wheel and really helped to improve my times.
I am very happy with the result i am getting.