PDA

View Full Version : Baseline Wheel FFB Values - Google Sheet & FCM. Universal



Pages : [1] 2 3 4 5 6 7

poirqc
17-07-2015, 02:04
Project Cars Force Feedback Guide

The idea behind this guide is to answer questions that the official guide may have left unanswered. Most of the information here come from hands on test from forum users and how we interpreted the data. While there's some link to complete force feedback templates, the idea here is to adapt Project Cars force feedback tools to your wheel. For a while, i talked about linearity. But it's not telling the whole story. I think we're more talking about Output Optimisation. So even if the sheet / FCM may suggest values that tend to be linear, the end goal is to modulate the FFB signal to make your wheel behave the best it can.

Test drives
This is the most important part.

Since patch 8.0, default settings are a good reference if you're lost at some point.

What i can recommend when doing test drives is to pick a track and a race car you know. You should have a easier time tracking how the force feedback react to curbs and road bumps. As for race cars, they tend to have a more direct link to the road. It's a good thing to do time trials. You won't lose time going out of the pits and the tires are already warm. I suggest you do 2 or 3 laps at once. Do a normal lap for the first one. Trow the car around for the others to see how it behave in tight situations. Change only a parameter at once. It will be easier to track how things changed. If you're to a point where it feels good. Keep the same track but try different kinds of car that you like. Touring, Road, LMPS. Mix them.

If they all feel good, congratulation. If not, take back you're first car and fix what driving behavior you want to adjusts with the corresponding tool. There's some small explanation below about witch does what.

Here's 5 place where you can get complete FFB templates:

pCars Feedback Settings (http://pcars.oscarolim.pt/)
Jack Spade'S suggested globals and car FFB (http://forum.projectcarsgame.com/showthread.php?22938-Jack-Spade-FFB-Tweaker-Files)
Grimeydog's can you feel it! (http://forum.projectcarsgame.com/showthread.php?38744-Lets-talk-FFB-Post-and-Compare-short-Telemetry-Videos-and-what-FFB-tweek-used&p=1111504&viewfull=1#post1111504)
bManic's informative one. (http://forum.projectcarsgame.com/showthread.php?25046-bManic-s-latest-FFB-experiments-possibly-working-universally-for-all-wheels&highlight=bmanic)
Yours truly G27 journey. (http://forum.projectcarsgame.com/showthread.php?34461-Yet-another-G27-FFB-template&p=1038028&viewfull=1#post1038028)
naondave's DFGT (http://forum.projectcarsgame.com/showthread.php?35147-Baseline-Wheel-FFB-Values-Google-Sheet-amp-FCM-Universal&p=1318907&viewfull=1#post1318907)
jpmmuc's TS-PC/T300 (http://forum.projectcarsgame.com/showthread.php?35147-Baseline-Wheel-FFB-Values-Google-Sheet-amp-FCM-Universal&p=1319267&viewfull=1#post1319267)

Keep in mind that the official guide (http://forum.projectcarsgame.com/showthread.php?29907-Force-Feedback-Guide-for-Project-CARS&p=969970&viewfull=1#post969970) is always a good reference if you think there's anything missing here.

224507233159
You can check the sheet here (https://docs.google.com/spreadsheets/d/19bx9hzTRa4HLqHtt1wgLfqxlN1PVHEv6rpTnm6jnZ6U/edit?usp=sharing),by Diluvian, use this sheet if you want to input your data. You can also try Force Curve Modifier (http://forum.projectcarsgame.com/showthread.php?35147-Baseline-FFB-values-Linear-FFB-Google-sheet-Universal&p=1110835&viewfull=1#post1110835) by skoader.

Link for Force Curve Modifier (http://forum.projectcarsgame.com/showthread.php?35147-Baseline-Wheel-FFB-Values-Google-Sheet-amp-FCM-Universal&p=1240034&viewfull=1#post1240034)

After lots of discussions, Diluvian wrote a sheet and skoader wrote a program! It will give you base values as a starting point. Right now, the mathematical model is pretty good, but it's not perfect. There's some Logitech Driving Force Pro data but those tools are made to work with any brand of wheel. Wheelcheck.exe will output the same data layout for every wheel model. You can import wheelchek data into Force Curve Modifier.

There's 2 things to keep in mind:

On a PC, as soon as you change any settings in the windows drivers, you'll get different data from wheelcheck / FCM
On Consoles, we don't really know how the wheel drivers behave beside overall force strenght.(Some wheels have memory that keeps windows settings)


Here's some examples:

Logitech MOMO racing(black) (https://docs.google.com/spreadsheets/d/1P1nQuhlWxgmc6GSQc1Xf3CvgpvCxyAtTA_ew5iggdGo/edit?usp=sharing)
Logitech G27 (https://docs.google.com/spreadsheets/d/14-Y7dvnLy74B9xhq-DvB1Q1Z72ERvFDczO6-wrb1lDA/edit?usp=sharing)
Logitech G29 (http://forum.projectcarsgame.com/showthread.php?35635-G29-Discussions-Plus-Support-and-Settings&p=1111216&viewfull=1#post1111216)
Thrustmaster TX Racing Wheel Ferrari 458 Italia Edition (http://forum.projectcarsgame.com/showthread.php?35147-Baseline-FFB-values-Linear-FFB-Google-sheet-Universal&p=1105092&viewfull=1#post1105092)
Accuforce Pro (http://forum.projectcarsgame.com/showthread.php?35147-Baseline-FFB-values-Linear-FFB-Google-sheet-Universal&p=1110627&viewfull=1#post1110627)
Fanatec GT3 RS & CSR (http://forum.projectcarsgame.com/showthread.php?22938-Jack-Spade-FFB-Tweaker-Files&p=1106058&viewfull=1#post1106058)
Fanatec CSW v2 (http://forum.projectcarsgame.com/showthread.php?22938-Jack-Spade-FFB-Tweaker-Files&p=1105406&viewfull=1#post1105406)
T150RS (https://docs.google.com/spreadsheets/d/1tWiu8uNOa-TjsvltxdYJCUkJrOvrywueAfgctY-BO9s/edit?usp=sharing)
T300RS (http://forum.projectcarsgame.com/showthread.php?35147-Baseline-FFB-values-Linear-FFB-Google-sheet-Universal&p=1110954&viewfull=1#post1110954)
T500RS (http://forum.projectcarsgame.com/showthread.php?35147-Baseline-controler-FFB-values-Google-sheet-amp-FCM-Universal&p=1143004&viewfull=1#post1143004)


Sheet usage:
0) Download 228642.
1) Click on the sheet link, let it open up, once the sheet is open go to MENU
2) FILE -> MAKE A COPY, name it whatever you want. You will need to have a google account to do this.
3) It will open a new sheet as a new window or new tab
4) You start your wheelcheck.exe and let it run 4x times on "Step Log 2 (linear force test)" wait until it finished, it will jump to "disabled" in the drop down menu
5) You go to folder "document". Open the first CSV with the editor and press CTRL+A "copy all" and go to the the workbook Log1 and paste CTRL+V into green field (just left click once and press CTRL+V) and you do the same, for the other 3x CSV files and LOG2, LOG3 and LOG4 workbooks in the sheet.
6) Once all that is done, automatically updates the CALCULATION workbook, you can look at it, screenshot it whatever you want.
7) Go to workbook PROJECT CARS and start adjusting your numbers(*) until your RED LINE gets as close to the YELLOW INPUT line! Take those values to your game settings and fine tweak and tune until you have it. that's it.

*Leave SG to 1, mess with it at the end, if at all. Put DRR at the start of the orange line, DRF between a 1/5th to a 1/10th of DRR. SK around the crossing of the shalow and step part. SR to lower the RED LINE.
**Quick tip (http://forum.projectcarsgame.com/showthread.php?35147-Baseline-FFB-values-Linear-FFB-Google-sheet-Universal&p=1113413&viewfull=1#post1113413), don't waste your time with values you could get from the min force test. Currently data from the "Step Log 2 (linear force test)" is better.
*** You don't have to run the line perfectly straight, at all cost. You'll have to drive the settings. From there it's probably better to read the tools description below and adjusts accordingly.
****Don't try to download and open the sheet in Excel or OpenOffice. It will open but the calculation tab won't translate and it won't work. It has to do how your system handle numbers with , or .

FCM usage:
Wheelcheck import:

Open FCM
Go into File --> Import --> go to your folder "Documents"
Select the .csv results wheelcheck generated. (You can import multiple files at once.)
Push the "Calculate Best R2 values" button.

FCM linear test: (http://forum.projectcarsgame.com/showthread.php?38744-Lets-talk-FFB-Post-and-Compare-short-Telemetry-Videos-and-what-FFB-tweek-used&p=1282826&viewfull=1#post1282826)

Open FCM
Go into Wheel --> Force Test
Pick between 2 and 4 samples. Let the step duration as it is. Set the DoR the same as your drivers.
Let run the test until it's done.
Push the "Calculate Best R2 values" button.


At that point, you may say: This isn't Project EXCEL, we're driving CARS here, what the hell all those number means?

First of all, this is a way to try to get a base configuration of a steering wheel. This is merely a starting point. Test drives will be mandatory. From there, you'll be able to make certains feelings come out more by fine tuning the different tools we have access to. I've included some tips how thoses tools works.


Terminology:
FFB: Force Feedback.
Torque: Applied force.
Delta: The sized of the difference between 2 things.
Slip angle: As you turn your wheel, they'll get more "grippy". At some point, you'll have less grip instead of more. You're past Optimal slip angle.
The 2 term below can mean alot of different things, based on context. In this guide, it mostly refers to:
Understeer: When the wheel gets light because you're past the optimal slip angle.
Oversteer: When you dive in a curve too hard and the rear end gets out. You have to counter-steer to catch the slide.
Oscillation: Small signal delta around the more nature FFB signal.
Wheel Acceleration can mean 2 different things, based on context. In this guide, it mostly refers to:

When you drive in a straight line, the car wheel's should align themselves when you let go the wheel, up to fairly high speeds. For various reasons, as soon as you let go the steering wheel, it can go left right left right more by itsefl, instead of slowing down.
When adjusting Damping, when PWM & PWMS are badly adjusted, turning the wheel won't be linear. The wheel movement will get progressively faster as you turn. This isn't wanted.
Torque Delta: A change of torque happening over time. It's either the FFB signal moving up or down. When the line is above or below 0, but is staying mostly flat, this is static torque.
TDC:Top Dead Center
FFB Deadzone:Part of the signal that is too low to make the wheel move. Usualy happend around TDC, but can be everywhere. It can also happend if the signal is too fast for the wheel to follow. You can see it in the telemtry but the wheel doesn't actually move.
Linear signal:Part of the signal that gets output as is, 1:1. It's strong enough to move the wheel, but low enough so it's not compressed.
Compressed signal:Part of the signal that gets compressed by RAC or Soft Clipping. The more you compress the signal, the smaller the torque delta will be.
Saturated signal: This can happend anywhere in the telemetry. Could happend in low forces with too much Scoop Reduction. Could happend near max torque when TF is too high against RAC or Soft Clipping.
Clipped signal:Well, easy enough, this signal goes above 1 so the wheel just output maximum torque.


Force Feedback Balance Plan
If you're using Jack's files (http://forum.projectcarsgame.com/showthread.php?22938-Jack-Spade-FFB-Tweaker-Files), install them correctly before playing with the controller FFB. It'll save you time because you won't need to tweak controller FFB before and after. At the same time, If you balance the Car FFB yourself, it will affect how much Tire Force is needed.
It's probably hard to plan how you'll balance the FFB of pCars. Depending on the wheel you have, you may reach a point where you'll have to focus on a mix of forces you want to feel. You may want something balanced towards subtle road feel, an overall stiff or soft steering, an emphasis on over/understeer. The key point is that it's probably better to have 2 nice force behaviors than 4 badly blended. Tennebaum's tennenbaum's take #1. (http://forum.projectcarsgame.com/showthread.php?38744-Lets-talk-FFB-Post-and-Compare-short-Telemetry-Videos-and-what-FFB-tweek-used&p=1145157&viewfull=1#post1145157) Here's #2 (http://forum.projectcarsgame.com/showthread.php?38744-Lets-talk-FFB-Post-and-Compare-short-Telemetry-Videos-and-what-FFB-tweek-used&p=1247377&viewfull=1#post1247377)

There's 3 big group of forces. Those that comes from the front(Fx,Fy,Fz & Mz) and rear wheels(SoPs). The 3rd group comes from G-Forces(Bodys in Car FFB) when you trow your car into a curve. You can see how things interact with each others. This come from the official guide. Credit to whoever drawn it:
224509
The balance between those 3 group can be vital. Those are the 3 chains of physics input that feed the FFB. I'll explain more about Tire Force below. Here's also a nice post about how those force translate into driving behaviors (http://forum.projectcarsgame.com/showthread.php?22938-Jack-Spade-FFB-Tweaker-Files&p=1226183&viewfull=1#post1226183).


Tire Force
This is the overall multiplier of the Car FFB balance. The Car FFB can be set via the UI or tweaker files (http://forum.projectcarsgame.com/showthread.php?22938-Jack-Spade-FFB-Tweaker-Files). I won't go into details about how to balance that part. It's a whole subject by itself. What matters here is the overall max forces those settings can dish out. An easy test that is to enter a curve really fast, trail break hard without locking the breaks. Try to get to max slip angle. It should reach the top of the telemetry hud. If it reach it and the yellow line gets flat, you're clipping. TF is too high. If the FFB signal gets really close but it keeps moving,without going flat, you're set.

When playing with Tire Force, from a low value, going up, the following happends:

The FFB can be linear, there's no compression going on. It give a lightweigh wheel, on most hardware, since their max torque output is not really strong. It could be stronger on a Direct Drive wheel.
As you move TF(For simplicity sake) up you start to need to compress the end of the torque slope to prevent the signal from being clipped. At that point, the beginning of the slope is linear, and the end is compressed.
At some point, the signal may not clip, depending on Relative Adjusts or Soft Clipping, but the FFB signal gets saturated. From my point of view, a saturated signal is one where the torque deltas are too hard to discern.
Finaly, if there's too much TF against RAC or Soft Clipping, the signal gets clipped too often.

In essense, the higher TF is, the harder anti-clipping(RAs or SC) tools will have to work. Tennenbaum did an hands on test, videos included, about what TF does against Anti clipping tools (http://forum.projectcarsgame.com/showthread.php?38744-Lets-talk-FFB-Post-and-Compare-short-Telemetry-Videos-and-what-FFB-tweek-used&p=1274304#post1274304).

The last thing i could say about TF is that when you run a setup with only FxyzMz, the Spindle Master Scale(Car FFB) and TF are swappable. As soon as you use any SoP, It's better to use TF instead of moving both SoP Scale and SMS. TF will move the FFB of everycar up or down where SMS and SoP Scale will only do it per car.

tennebaum wrote another great post (http://forum.projectcarsgame.com/showthread.php?38744-Lets-talk-FFB-Post-and-Compare-short-Telemetry-Videos-and-what-FFB-tweek-used&p=1204620&viewfull=1#post1204620) about how the Car FFB behave againt Tire Force.


PWM / PWMS / WPS(Damping)
FFB wheel construction is very different depending on your wheel's model. When they're not plugged, some will be very free moving and other will be dragged by the wheel's motor. This tool is used to compensate for that. You'll be able to set the behavior of the steering wheel to you taste with that. You want turn ins and turn outs to feel natural to you, where subconsciently, when you turn, it turns at that moment, to the pace you want. Not before, not after.

I'll talk about Per Wheel Movement, Per Wheel Movement Squared and Wheel Position Smoothing as PWM /PWMS / WPS

There's 3 cases where you want to use this tool:


You like the build in drag of the wheel, it's turning as expected
There's some "lag" when you turn, you want more responsiveness
It's too twitchy when you turn, you need to correct your line too quickly

PWM 0/PWMS 0 won't change the default behavior of your steering wheel
PWM positive value/ PWMS negative nalue will give you a slower moving wheel
PWM negative value/ PWMS positive value will give you a faster moving wheel

In both case, you don't want turn ins and turn outs that accelerate. You'll understand if you mess with those values. Most of the time, low values will do the trick.

This is where WPS comes in. Depending on you're whole controller FFB setup, you'll probably need some WPS. The more you put, the less the wheel will oscillate by itself. However, you may lose low torque details at some point. Idealy, you need to find a balance where the wheel doesn't oscillate too much while keeping as much details a possible. Depending on how much smoothing you use within the Car FFB, you may need less WPS because of that.


Deadzone Removals(TightenCenterRange/Falloff)
A common agreement is that every wheel force feedback motor has a deadzone. A place where it can't generate smaller forces. Some wheel have lots of it. Others don't. Without any calculation you'll be able to see how much of a deadzone a wheel has. In pCars, this is dealt with Deadzone Removals.

If you're using the sheet, the following can apply:
DRR = Deadzone Removal Range : You'll set this value where the orange line(Normalized wheel output) starts. You may move that value a small bit(more or less 0.01).
DRF = Deadzone Removal Falloff : Start with something like a 1/10th of DRR and try other values so the red line(Final wheel output) follow the yellow line(input).
** About the DRF, the ui doesn't show 0,00X increment. They are taken into account tho. You can move the value with the arrows.

If we check how Deadzone Removals translate to driving, DDR would be how strong the signal gets boosted around TDC. DRF is how fast the signal boosted to DRR value. Depending on your Soft Clipping usage or high TF value, you may not need to use much Deadzone Removals, if at all. At the other end of the spectrum, if you have a setup who doesn't boost low forces from other settings, a DRR/DRF combo will be needed.

As soon as you use Deadzone Removal Range, you must use Deadzone Removal Falloff. Those 2 settings are tied together. If you don't have enough DRF, you'll get a squared wave osciallation when going in a straight line. An easy way to tune DRR/DRF is to set DRR like mentioned above, DRF would be a 1/10th of DRR. After this is done, play with DRF until force ramp up quickly enough, without a strong notch ,on turn ins.

Just did some quick test about wheel oscillation and acceleration.

When someone picks a given DRR, this is what happends with various DRF:

No DRF: Square Wave oscillation around center. It can tattle. (It's also easy to see in the telemetry hud.)
Not enough DRF: The square wave will go away, but you can feel a notch when turning from left to right quickly.
Enough DRF: The centre of the wheel feels right, no notch when turning from left to right quickly. It can help a bit with some rattling. The wheel should self align by itself when going straight, up to some great speed.
Too much DRF: The wheel will start to accelerate, before you go really fast, just by moving it a tiny bit and letting it go. Depending on your globals, it may not give enough forces on turn ins.



Relatives AdjustsNeed some cleanup and precisions
Like Soft Clippings, Relatives Adjusts can help you deal with clipping. It's a signal shaping tool. There's lot of ways of using it. You could check tpw's post (http://forum.projectcarsgame.com/showthread.php?34527-FFB-Experiments-with-Relative-Adjust-parameters&p=1038876&viewfull=1#post1038876), the official guide (http://forum.projectcarsgame.com/showthread.php?29907-Force-Feedback-Guide-for-Project-CARS). I'll explain them briefly. In pCars, they don't come in that order but i think it makes sense that way. You can also disable Relatives Adjusts by putting the 3 RA's at 0.

There's one thing to keep in mind when messing with Relative Adjusts. It's easier to manage without Soft Clipping. If you add Soft Clipping afterward, you'll probably have to come back at RAs to align both tools. They need to be aligned to work great. Another thing is that RAs get his input only from the wheels. The G-Forces are applied afterwards.

Relative Adjust Gain: This parameter will let you control how much you want to lower or boost the amplitude of a FFB torque delta. You could say you "spice" or "tame" the signal.
Relative Adjust Bleed: This is how long the torque variation will decay, or bleed. If you use Relatives Adjuts, don't use a 0 value. The wheel will decenter itself.
Relative Adjust Clamp: This is the value where the FFB will oscillate around when RAs prevent clipping.

Work in progress:You shouldn't even read that part, i'm thinking out loud! :D
I deal with RAs this way:

Find the right offset between RAG and RAC
Adjust RAB until weight transfers feels good
Lower RAG / RAC by the same amount to prevent clipping that may occurs from moving RAB up from it's default 0,10 value.


Scoops
Scoops is tool to change the balance of weak vs strong forces. The best way to see how Scoops works is by playing with the FCM application. You'll see it for yourself! Sometime easy to feel is that, by itself, Scoop Reduction will change how much road feel you get, when going straight. You don't have to move it much to get the results you want.

Obviously, Scoops aren't limited to specific steering wheel behaviors. You do what you want with it. However, the following 2 go hand in hand. If we translate all that to driving, Scoops can be used to help you find the balance between how fast tire load ramp up and how much understeering feeling you want to get. The Stronger the wheel is, the smaller the available torque delta, to feel understeering, will be.

To quote Haiden (http://forum.projectcarsgame.com/showthread.php?38744-Lets-talk-FFB-Post-and-Compare-short-Telemetry-Videos-and-what-FFB-tweek-used&p=1277487&viewfull=1#post1277487):

With that knee, SR=0.20 provides the best feel, immersion wise. It's a very firm wheel with good road feel, and has the best cornering dynamic (with regard to range of increasing force). On the other end, SR=0.28 sacrifices some firmness and road feel for more pronounced/detectable slip feel.

Time up! I'll happen the understeering feeling later when i get some time.

Without moving the knee:
More Scoop Reduction = Less road feel
Less Scoop Reduction = More road feel

Without moving the reduction:
More Scoop Knee = More road feel
Less Scoop Knee = Less road feel

Ref, will explain later. (http://forum.projectcarsgame.com/showthread.php?38744-Lets-talk-FFB-Post-and-Compare-short-Telemetry-Videos-and-what-FFB-tweek-used&p=1277651&viewfull=1#post1277651)


Soft Clippings: Skoader updated the FCM with SC support, need to rewrite this part.
I can say is that it's main use is to prevent clipping. It is also a signal shaping tool. Tennebaum wrote another great post (http://forum.projectcarsgame.com/showthread.php?38744-Lets-talk-FFB-Post-and-Compare-short-Telemetry-Videos-and-what-FFB-tweek-used&p=1213005&viewfull=1#post1213005) on Soft Clippings. It's the last step of the signal chain that can be seen as "generic". It's not tied to any wheel. It process the overall combination of the inputs. (http://forum.projectcarsgame.com/showthread.php?38744-Lets-talk-FFB-Post-and-Compare-short-Telemetry-Videos-and-what-FFB-tweek-used&p=1205115&viewfull=1#post1205115)

Ref, will update later (http://forum.projectcarsgame.com/showthread.php?38744-Lets-talk-FFB-Post-and-Compare-short-Telemetry-Videos-and-what-FFB-tweek-used&p=1286357&viewfull=1#post1286357)

If you want to actually put more TF into the mix, you'll need Soft Clipping. SC act as a signal compressor to lower total forces into the usable range of pCars FFB. One thing i can say is that Soft Clipping Half Input:1 and Soft Clipping Full 2.09, values from the official guide, works fairly well on low end wheels and probably most wheels. Depending on your Soft Clipping settings, you may need to move Deadzone Removals and Scoop a bit.

The first step is to move SCFO(with SCHI half of SCFO) up until you're not clipping under heavy wheels load.

When SCFO is dialed to "catch" everything:

SCHI < 0,5*(SCFO) = Expantion of the low forces, less "space" for the high forces
SCHI > 0,5*(SCFO) = Compression of the low forces, more "space" for the high forces

There's one thing to keep in mind when messing with Soft Clipping. It's easier to manage without Relative Adjusts . If you add Relative Adjusts afterward, you'll probably have to come back at SC to align both tools. They need to be aligned to work great.


Steering Gain(Work in progress, take it with a grain of salt)
By design, Steering Gain is the final step of the telemetry signal chain. It acts as a volume control of the mixed FFB signal. Here's the official words on it:

Steering Gain – the gain (multiplier) applied to all steering effects (steering force, jolts, kerb rumble etc) after they have been mixed. For a clean more detailed experience set at 1.0 or below, for stronger feedback at the expense of clipping set higher (maximum value 5).

There's 3 use case for this tool:

It lowers the overall forces a wheel produce.
When SG is at 1, the signal doesn't change.
It boosts the forces a wheel produce.

One thing to keep in mind about SG is that it won't prevent clipping that is already happening. It can also cause clipping if RAs or SC are not used in tandem with SG.

Lowering the signal:
The easiest way to use Steering Gain is to lower the FFB signal. Some wheels are can dish out strong forces. At some point, someone may want to soften that and Steering Gain will do that. Keep in mind that you should lower the FFB by the least amount you can. As soon as you lower the FFB range, more of signal will be in the FFB deadzone of a wheel. For low deadzone wheel, it may not be a problem. But for high deadzone wheel, it will be felt.

This means that if you plan to use SG < 1, you'll have to recalculate Scoops and Deadzones (http://forum.projectcarsgame.com/showthread.php?35147-Baseline-Wheel-FFB-Values-Google-Sheet-amp-FCM-Universal&p=1182729&viewfull=1#post1182729).

Leaving the signal unchanged:
Steering Gain = 1. This is self explanatory.

Boosting the signal:
There's a couple of ways to boost the force feedback strenght. If we focus on SG, it will boost the whole range of forces. This means you can use it to get that last extra space of forces. Under most situations, if the only thing approaching max range are the FFB spikes, you can move SG up slowly. It's important to avoid clipping important informatives forces. Chopping the top of spikes isn't that bad.


Master FFB(Controls --> configuration --> Force Feedback)
There was some debate, for a while, about the correct usage of this parameters.

It seems to work differently for the PC and Consoles...

On the PC version of pCars, this setting is redundant. Just set it to 100 and use the windows control panel to control your steering wheel strenght ouput.
On the console versions, a safe bet is to leave the strenght to its default value. There has been report of damaged wheels when set to 100%. For Thrusmasters wheels, the manufacturer recommends to leave it to default value(Link needed). However, some users where able to use 100%, depending on their whole globals settings. On consoles, it's essentially the driver's control panel of the wheel. The only exceptions are the fanatec wheels. They have built-in memory that keeps the settings you set in the windows control panel.

In either case, when you lower the maximum torque a wheel can dish out, It'll change the natural output of the wheel. It will also increase the FFB deadzone of the wheel.

Intersting tip
Skoader came up with a nice trick to go around this problem! Instead of lowering the game FFB, you set Scoop Knee to 1 and Scoop Reduction to the amount you want to lower the max output. That way, the max output gets lower without actually increasing the FFB deadzone!

Controller sensitivity & deadzones
In the Controls / configuration tab, you can find couple settings about deadzone and sensitivity. We're talking about the input of our wheels. It's how the sensors in our wheel/pedals will behave.

Deadzones are there so that worn sensors don't give jittery data. If your hardware is in good shape, you probably dont need much deadzones, if at all.
Sensitivity serves a different purpose. First of all, let's talk about how it works. 50, the default value, mean that the range from 0 to 1 is linear. As you move the value toward 0, you'll get more precision at the beginning of a given input. This will lead into less precision at the end of that input. What does that mean for the wheels and pedals.

Wheel: For any wheel that has 900 or 1080 DoR, it's better to leave it at 50. You already have enough precision for thoses small movements, and you don't want less precision as you turn. However, you may need to lower this value if you have a wheel that has less than 900 DoR. Let's take a momo wheel for example. It only has 240(270, i don't remember) DoR. Since we rarely go full lock while racing, it may be usefull to lower the sensitivity, you'll get more precision on turn ins. The wheel will feel more natural top dead center.
Throttle: Since we're racing, most of the time, we have this pedal to the floor! With this in mind, it make sense to increment above 50. If not overdone, this will give you precision when going towards full throttle.
Brakes: Lot's of time, you'll need to brake just a tiny bit. Lowering the sensitivity for brakes, a bit below 50 will give you precision when doing feathering the brakes.
Cluth: This pedal will be fully pressed when it matters, so a value a bit above 50 will make stoped start a bit easier. You'll be able to more precisely catch the point of friction when doing a stopped start on a race.



If you get to a point where you feel you've got a solid FFB setup, feel free to post how you configured Project Cars for your wheel!

I want to give credits to all of those that came up with the stuff needed to build it. Diluvian, skoader, VictoriSV, Schnizz58, PTG Ty1er Ward, tpw, STEELJOCKEY, dw123, Jack Spade, rocafella1978, GrimeyDog, tennenbaum, Haiden, morpwr, and everyone who's adding their views on the subject!

Here's a link to Grimeydog's FFB megathread (http://forum.projectcarsgame.com/showthread.php?38744-Lets-talk-FFB-Post-and-Compare-short-Telemetry-Videos-and-what-FFB-tweek-used). It's long read, but there's talk about all aspect of pCars FFB.

As always, if anyone think something is wrong in this post, speak your mind!

Enjoy! :D


Update #23/05/2016:Added some bits about Steering Gain and Game Force Feedback.
Update #21/05/2016:Rewrote some part of Deadzone Removals.
Update #13/05/2016:Added some info to Tire Force. Started to update the Scoops refinition. Added link to complete FFB templates. Updated FCM screenshot. Various small stuff.
Update #24/02/2016:Updated FCM 1.2 links. You can get to the original post and the latest one about SC.
Update #11/02/2016:Added some link to good posts.
Update #23/01/2016:Put the test drive part on top because i think it's the final stamp of approval. Numbers don't mean anything if it doesn't drive well.
Update #30/12/2015:Updated the first steps to use the sheet. Will try to build a howto for FCM also.
Update #26/12/2015:I messed up the whole post and i had to rebuild it. There may be parts that where not properly formated.
Update #14/12/2015:Tire Force cleanup. Added various bits and a terminology part.
Update #07/12/2015:FFB Balance Plan & Tire Force
Update #XX:Soft Clipping and Relative Adjusts, alone or mixed.
Update #XX:Controller sensitibity and deadzone added.
Update #10:I'll try to update the post with a small tutorial to open and use it in google sheet.
Update #8:New sheet version available: Simple copy&paste wheelcheck logfile functionality
Update #7:Cleanup and RAs.
Update #6:Removed the link to my sheet and replaced it with Diluvian's one. Added some comments about FFB tools.
Update #5:You can now calculate Scoops fairly easily now! I'll be able to catch some sleep! images/smilies/biggrin.png
Update #3:Nothing is set in stone right now, some discussion are still going on and there's much to understand. We'll see how it develops after patch 3.0

b_akerlite
20-07-2015, 09:33
Nice job!

jhonatas
21-07-2015, 14:25
please can you be specific? I do not understand.

Import the .csv into the sheet, voila. ???

I do not understand almost anything excel if possible give me step by step.

poirqc
21-07-2015, 14:41
You need to have access to a PC to hook up the wheel and do a linear test in wheelcheck.

If you don't have access to microsoft excel, you can install open office or create a google account and use google sheet. There's should be plenty of guides for the on the web.

After that, you open the(whatever format) i gave in the first post. You'll have the data i have for my wheel. You can remplace that data by importing the data you collected by doing the wheelcheck test. If you replace it correctly, it'll get graphed automaticaly.

Hope this helps.

jhonatas
21-07-2015, 14:55
thanks.

poirqc
25-07-2015, 13:06
Well, first post updated.

inthebagbud
26-07-2015, 09:22
Ok I am going to ask the question, what does the graph actually show me

Lars Rosenquist
26-07-2015, 09:29
Ok I am going to ask the question, what does the graph actually show me

It shows the physical response of the wheel in response to a value that is output by the software. So the game tells the wheel 'move 10%', then the wheel responds with a certain motion (e.g. 8%). If response would be lineair, the wheel would always respond the way the game tells it to, but this is rarely the case, unless you have a really expensive wheel. This graph makes it possible to configure the FFB to compensate for this behavior.

Does this help you?

Edit: I'll run it with my G27 tonight too. Oh how I wish games/sims would incorporate this and create a profile for your wheel automatically. :)
Edit2: Can you also add your LT profiler settings, since these (especially damping) affect the output if I recall correctly.

inthebagbud
26-07-2015, 09:52
It shows the physical response of the wheel in response to a value that is output by the software. So the game tells the wheel 'move 10%', then the wheel responds with a certain motion (e.g. 8%). If response would be lineair, the wheel would always respond the way the game tells it to, but this is rarely the case, unless you have a really expensive wheel. This graph makes it possible to configure the FFB to compensate for this behavior.

Does this help you?

Edit: I'll run it with my G27 tonight too. Oh how I wish games/sims would incorporate this and create a profile for your wheel automatically. :)
Edit2: Can you also add your LT profiler settings, since these (especially damping) affect the output if I recall correctly.

So in your graph your ddr is around 16?

Does this checker only work for PC/wheel combinations and if not where would you download the wheelchecker file

poirqc
26-07-2015, 12:00
So in your graph your ddr is around 16?

Does this checker only work for PC/wheel combinations and if not where would you download the wheelchecker file

Yes, Wheelcheck (https://mega.co.nz/#!wt4UzboQ!r70tMEAqOxY1UZxSviiukJs29OfSyIBBJYULPWWyNCg) works with any wheel.

Thanks to Photonenbert for posting it.

Lars Rosenquist
26-07-2015, 12:30
So in your graph your ddr is around 16?Yes, for mine it is. Values can vary per model and even per unit though. But you'll usually see roughly similar values per model. One of the exceptions there is the DFGT which varies wildy.

poirqc
26-07-2015, 12:59
Yes, for mine it is. Values can vary per model and even per unit though. But you'll usually see roughly similar values per model. One of the exceptions there is the DFGT which varies wildy.

I wonder if it varies from the min force test or from the linear force test. I would be good if people posted their results here.

inthebagbud
26-07-2015, 14:41
poirqc

Sorry for not understanding fully the process

in your sheet you have 4 tabs and the graph is based on the average of the 4 - are they just 4 separate tests

my initial screen looks like this

213359

and gives the following results

force startX endX deltaX deltaXDeg
0 0 0 0 0
200 -37 -37 0 0
400 -37 -34 3 0.178531
600 -34 -24 10 0.425727
800 -25 -21 4 0.164797
1000 -22 -14 8 0.370794
1200 -17 268 285 12.840466
1400 68 471 403 18.127718
1600 42 660 618 27.782101
1800 -26 878 904 40.7187
2000 -34 1203 1237 55.742733
2200 -42 1359 1401 63.089951
2400 -76 1584 1660 74.763107
2600 -42 1905 1947 87.672234
2800 -21 2132 2153 96.955826
3000 -19 2230 2249 101.254295
3200 -18 2464 2482 111.760132
3400 -15 2833 2848 128.212402
3600 -16 2942 2958 133.12886
3800 4 3198 3194 143.730835
4000 -3 3326 3329 149.814606
4200 -2 3583 3585 161.377884
4400 -10 3665 3675 165.429153
4600 0 3761 3761 169.301895
4800 19 3971 3952 177.83017
5000 4 4025 4021 180.961319
5200 19 4082 4063 182.856491
5400 37 4114 4077 183.460739
5600 26 4234 4208 189.352249
5800 48 4259 4211 189.530792
6000 31 4250 4219 189.874115
6200 22 4317 4295 193.307388
6400 14 4361 4347 195.587097
6600 19 4360 4341 195.326157
6800 9 4341 4332 194.969101
7000 5 4352 4347 195.642029
7200 21 4368 4347 195.642029
7400 29 4451 4422 198.979172
7600 35 4511 4476 201.409927
7800 23 4530 4507 202.824448
8000 13 4472 4459 200.654617
8200 -1 4481 4482 201.766998
8400 9 4549 4540 204.280151
8600 14 4558 4544 204.458679
8800 13 4566 4553 204.884415
9000 13 4502 4489 202.014191
9200 24 4562 4538 204.211487
9400 -12 4535 4547 204.650955
9600 -10 4563 4573 205.818268
9800 12 4601 4589 206.49118
10000 9 4534 4525 203.607239

poirqc
26-07-2015, 15:01
inthebagbud, it would be easier for me to paste the data in the sheet if you attached the .CSV in the post instead. I have a hard time parsing the data when i copy paste it from some bare text.

inthebagbud
26-07-2015, 15:08
inthebagbud, it would be easier for me to paste the data in the sheet if you attached the .CSV in the post instead. I have a hard time parsing the data when i copy paste it from some bare text.

sorry for that - it didn't like upload of a csv so here is a zip with 4 output files 213365

Thank you in advance for your time

poirqc
26-07-2015, 16:31
sorry for that - it didn't like upload of a csv so here is a zip with 4 output files 213365

Thank you in advance for your time

There you go

213375

I attached the graph in 2 formats. You should be able to import them in google sheet, Excel or Open Office.

inthebagbud
26-07-2015, 18:22
poirqc

Thanks for the file, which gives me a graph like this

213391

I would like to understand some of the information you have used from the table so I can interpret for myself what is being said


lineal step -this is derived from cell L51/45 - is this because cell L51 is where output = 100 and is divided by 45 because there are 5 steps before the mean value rises above the value of 1 (50 steps minus 5 = 45)

Scoop Knee- is the output figure of 48 derived by judging on the graph where the knee starts which equates to cell K26

Scoop Reduction with/without DDR - this is where I struggle as you use cells L21 and M21 in the formula, but should this not be row 26 to match that of the scoop knee row value?

Interpreting the required DDR I assume this will be somewhere between 0.1 and 0.12 as this is where the output rises above the value of 1

213392

For anybody else using a TX wheel reading this please note my TM control panel settings are

213393

poirqc thanks again for taking the time to go through this

b_akerlite
26-07-2015, 18:51
Thanks for this - its nice to have another tool to help dial in the wheel.
One question - The scoop knee calculation in the spreadsheet is based on formula "k21/100" - this seems weird as k21 is always going to be 38 no matter what data you input?? so whatever wheel gets tested it will always show 0.38 for scoop knee?

also what does DRR stand for?

edit - ahh ok - so you have to visually and look for the kink in your blue line in the graph?
This makes sense - (as i use compressors for audio so I should have copped the knee earlier!)

So in my graph my blue line kinks at input value of 28 so I have to look for 28 which is K16 and change all the scoop calculation formulas to reference line 16.... is that correct?!!!

:confused:

Lars Rosenquist
26-07-2015, 20:34
Here's the output for my G27. Unsurprisingly, it's very similar to the G29 (minus wear and tear on my end I guess). :) I added the .ods in there as well as a zip file.

I screwed up the horizontal scale, but DRR is at 0.16 (16%)

213400

213401

Schnizz58
26-07-2015, 20:56
How do you run the test? I have selected Step Log 2 from the combobox but I don't see a button or anything to perform the test. I tried Refresh but that didn't seem to do anything.

inthebagbud
26-07-2015, 21:06
How do you run the test? I have selected Step Log 2 from the combobox but I don't see a button or anything to perform the test. I tried Refresh but that didn't seem to do anything.

Schnizz58 - wondered how long it would take until you appeared on here :)

make sure wheel is connected to pc and then run the software and select the step and it should auto start

Schnizz58
26-07-2015, 21:07
It doesn't do anything.

inthebagbud
26-07-2015, 21:09
It doesn't do anything.

daft question is wheel plugged into power socket as well?

Schnizz58
26-07-2015, 21:12
Yes, it had to talk to the wheel in order to fill out the status box on the right (and include it in the list of devices at the top).

Schnizz58
26-07-2015, 21:17
wondered how long it would take until you appeared on here
You know I can't resist the siren's song.

inthebagbud
26-07-2015, 21:21
Yes, it had to talk to the wheel in order to fill out the status box on the right (and include it in the list of devices at the top).

h'mm not sure why it is not working just tried again on mine connected to pc launched software, wheel is selected automatically and pick log 2 from list and after a couple of seconds wheel starts going through test routine

Schnizz58
26-07-2015, 21:25
OK well I'mma go watch the USCC from Lime Rock, lol. Peace...

ETA: Well I've never seen that before. There was an incident that took out the leaders in both classes at the same time!

poirqc
27-07-2015, 00:33
Well, i kept the Scoops calculation in the sheet. However, i still have too many question about it so take it with a grain of salt.

With that said, you can clearly see the deadzone of various wheels.

poirqc
27-07-2015, 00:40
Here's the output for my G27. Unsurprisingly, it's very similar to the G29 (minus wear and tear on my end I guess). :) I added the .ods in there as well as a zip file.

I screwed up the horizontal scale, but DRR is at 0.16 (16%)

213400

213401

Normally, if you replaced the data in the 4 sheets, the scale of the graph should stay the same. I wonder why it did that.

Have a look: 213421

Edit - first post updated

Lars Rosenquist
27-07-2015, 06:36
Normally, if you replaced the data in the 4 sheets, the scale of the graph should stay the same. I wonder why it did that.Probably a conversion issue because I opened the .xls in OpenOffice (so not the .ods you sent) and then saved it to .ods.

Flihp
27-07-2015, 11:09
here's mine, i'm pretty clueless what to make of it.

213454


Phil

jhonatas
27-07-2015, 12:46
I did minforce test with the Thrustmaster T300 and 0.05 presented the result, then it would DRR = 0 (which is already default), what value for DRF?

poirqc
27-07-2015, 13:04
here's mine, i'm pretty clueless what to make of it.

213454


Phil

It seems your DRR is about at 13% so 0.13 DRR. And you want a linear fell, you could disable Scoops because it's already pretty linear.

poirqc
27-07-2015, 13:06
I did minforce test with the Thrustmaster T300 and 0.05 presented the result, then it would DRR = 0 (which is already default), what value for DRF?

Well, you would have 2 things to do. Do multi samples of the mininum force test. To be sure the 0.05 value is good. Or you could do a Force linear test. It will probably show another DRR. Force linear tests seems to be more consistent value wise.

b_akerlite
27-07-2015, 13:33
From what I can make of it it tells us how linear the response of your ffb in the wheel is.
The orange line is an example of what exactly linear response would be so the software sends a signal of a value of 10 for example and the output force given by the wheel has a value of 10 and son on.
The blue line is Exactly what your wheel puts out so for example you can see with a software input force of 28 your wheel actually gives an output force of 20. And with a software input force of 46 your wheel outputs 40. And so on.

You can also see from your graph that it is not until an input force of 14 is given does your wheel give any actual output force at all so I imagine this would be the dead zone.

I thought I understood the scoop knee part of it and the DRR and how it is calculated from the data but now I am not so sure and REALLY hoping someone will chime in and explain!

Flihp
27-07-2015, 13:38
It seems your DRR is about at 13% so 0.13 DRR. And you want a linear fell, you could disable Scoops because it's already pretty linear.

Thx, I have my deadzone removal set to 0.11, and believe it or not I don't use any scoop settings, both on zero. I can put those two sliders anywhere and doesn't feel any different.

Phil

mazinga
27-07-2015, 14:54
My T300RS does not have knee, just belly.

213476

b_akerlite
27-07-2015, 15:23
My T300RS does not have knee, just belly.


then i definately musunderstand scoop kneee as it looks to me like your knee is at 26 input / 30 output

edit - i edited your graph to show what I interpret the knee to be....

poirqc
27-07-2015, 16:01
Thx, I have my deadzone removal set to 0.11, and believe it or not I don't use any scoop settings, both on zero. I can put those two sliders anywhere and doesn't feel any different.

Phil

Well, i think it's a perfectly valid to not use Scoops. It can be a matter of taste, to an extent. If you like how you're wheel is responding, then, you just don't use the tool.

Some wheels are almost linear (G25, GT2/3), some have a bent or a knee(G27/G29), others seems to have a belly(T300RS). From there it will dependson how to use it. I think there's 3 way to use the tool:


Not use it at all - good
Use it with the correct amount of reduction at the knee - good
Useit with too much reduction at the Knee - not desirable


The 2 first scenario are good. Idealy, we should avoid the 3rd one because it will remove some information from the FFB signal


There's a nice discussion going on in the G29 thread (http://forum.projectcarsgame.com/showthread.php?35635-G29-Discussions-and-Support), PS4 side, i invite everyone to have a look. I've yet to process everything from there.

Cheers!

Schnizz58
27-07-2015, 16:12
then i definately musunderstand scoop kneee as it looks to me like your knee is at 26 input / 30 output

edit - i edited your graph to show what I interpret the knee to be....
I think the knee is higher...like around 56-60. See how the curve flattens out above that? You apply scoop reduction to make that more linear.

inthebagbud
27-07-2015, 18:01
Poirgc

I have had a look at the g29 thread and will need to study it a bit more to fully understand all that is being said.

However in connection with scoop knee and reduction are you trying to identify the reduction needed to bring the knee more in line with the linear outputs and if so is the idea to have as linear line as possible?

Looking at the graph I posted for the tx it is so far away from being linear than any of the other graphs posted here or in the g29 thread

jhonatas
27-07-2015, 18:23
My T300RS does not have knee, just belly.

213476

According to this chart then the DRR for the T300 should be 0.12?

jhonatas
27-07-2015, 23:20
You need to have access to a PC to hook up the wheel and do a linear test in wheelcheck.

If you don't have access to microsoft excel, you can install open office or create a google account and use google sheet. There's should be plenty of guides for the on the web.

After that, you open the(whatever format) i gave in the first post. You'll have the data i have for my wheel. You can remplace that data by importing the data you collected by doing the wheelcheck test. If you replace it correctly, it'll get graphed automaticaly.

Hope this helps.

how should I proceed to replace the data? I have done the test and already have log2 file, but could not import to excel.

jhonatas
27-07-2015, 23:45
in your file there are 4 spreadsheets, 100-1, 100-2, 100-3 and 100-4, the graph is an average of these values, but I could not get the 4 spreadsheets, what should I do? in my file has only one sheet.

however, I tried to do otherwise, I created a chart with 0-100 and deltaX scale and got the following chart

which can be interpreted by this graph?

213582

DRR 7% = 0,07?

poirqc
28-07-2015, 00:28
Poirgc
However in connection with scoop knee and reduction are you trying to identify the reduction needed to bring the knee more in line with the linear outputs and if so is the idea to have as linear line as possible?


Well, my train of thoughts behind all of this is to target that linearity as close as possible, as a starting point. In my view, this is like saying: Ok, my wheel can output the best range it can.

Afterwards, you would then tweak those values just a bit, to your driving taste. Because after all, this isn't Project Linearity, it's Project Cars.

From my testing experiments, most of the time, you don't have to move values much to target the feel you want. When you have a solid base to start from.

poirqc
28-07-2015, 00:31
how should I proceed to replace the data? I have done the test and already have log2 file, but could not import to excel.

I guess you're using excel?

If so, you should be able to open a .csv file. The easiest way should probably be to open the document i posted in Excel. After that, you open the .csv log2 files into a new document. From there, you'll be able to copy paste the data from sheet to sheet.

You just have to place everything in the same cells.

wadafaka
28-07-2015, 11:10
I did one run with my T300RS and this is the result:
213638

So, is that linear enough? I only need to set Deadzone removal?

--
By the way, for those who don't know how to import CSV easily into Excel:
http://superuser.com/questions/407082/easiest-way-to-open-csv-with-commas-in-excel#answer-407085

inthebagbud
28-07-2015, 12:04
Well, my train of thoughts behind all of this is to target that linearity as close as possible, as a starting point. In my view, this is like saying: Ok, my wheel can output the best range it can.

Afterwards, you would then tweak those values just a bit, to your driving taste. Because after all, this isn't Project Linearity, it's Project Cars.

From my testing experiments, most of the time, you don't have to move values much to target the feel you want. When you have a solid base to start from.

Ok so am i correct in that

1. if I interpret my graph below it has an "inbuilt knee" at around 48

2. if I have knee and reduction in the game set to zero then the wheel conforms to this in built knee

If these assumptions are correct then the question is how do we move the output line to become more linear

Also how can we check how the alterations to knee and reduction have impacted the curve as we cannot change this in game and then see the change in the wheelchecker

Maybe it should be done through the TX control panel?





213641

poirqc
28-07-2015, 12:35
@wadafaka :

Yeah, it seems pretty linear to me.
At that point, you could still use some Scoops to shape the FFB to your liking.

@inthebagbud :

#1: About that, yeah
#2: Yes

You won't be able to change the shape of the wheel response via the wheel control panel. Changing the strenght from there will only change the physical deadzone of the wheel. I sampled my G27 from 90% to 107%, with 6 different strenght values. As you put more strenght, the deadzone get's smaller, but the response keep the same shape(somewhat the same).

The best you could do it do it yourself. Easily, you could just do 1 sample per strenght. Then add those to the graph. You would see how it behave.

Schnizz58
30-07-2015, 21:19
I was able to get this to run on a different computer after some rebooting and futzing around. I have 5 columns in my spreadsheet, force, startx, endx, deltax and deltaxdeg. What do these 5 parameters represent? Which two are you guys using for your graphs?

inthebagbud
30-07-2015, 21:38
Schnizz58

It's a bit late and I have been for a few scoops so hope this makes sense .

Download the spreadsheet from op

Run the test 4 times and this will give you 4 different sets of information to use as an average In the graph

In the spreadsheet there are tabs 100-1 -2 -3 -4 cut and paste your info from 1st file into tab 1 2nd file into tab and so on for rest of info and this will update spreadsheet

It will be interesting to see your results as I have an opinion and need a second

Will be around again Saturday but not sure of time difference if you need further help

Lars Rosenquist
31-07-2015, 08:53
I was able to get this to run on a different computer after some rebooting and futzing around. I have 5 columns in my spreadsheet, force, startx, endx, deltax and deltaxdeg. What do these 5 parameters represent? Which two are you guys using for your graphs?

All of them. :) You must paste all the values in the sheets two through five, not in the first one (that shows the graph). The first one aggregates automatically from the next four. So run the program four times and paste the data in the appropriate sheet.

poirqc
31-07-2015, 13:30
If i'm not mistaken, you could run the test only 1 time and just delete the data in the 3 other tab (100-2, 100-3, 100-4). You just won't get smoother samples. But it should still work.

Schnizz58
31-07-2015, 14:17
OK thanks but what are the 5 parameters that the analysis program spits out?


It will be interesting to see your results as I have an opinion and need a second
Mine looks pretty much like yours.

The first thing I did was add in some deadzone removal. I set the range to .12 and the falloff is two clicks to the right (I think that's .06 but not sure). That definitely tightened up the center.

Then I started thinking about how to use scoop to make that curve more linear. That's an ongoing project but it also occurred to me that maybe the shape of that curve isn't so bad. The forces at the upper end of the curve where it flattens out are going to be things like impacts with the Armco and I don't particularly need that stuff to be modeled accurately. What I'm more interested in are the feel of the road surface at low forces and the feel of grip (or loss of it) when cornering (which I assume are going to be in the middle of the curve somewhere). So it's possible that the shape of the curve is okay. I've been driving with it that way for a couple of weeks and I think it feels pretty good. Nevertheless, I'm going to continue to play around with the spreadsheet to see how linear I can get it and then try those settings in the game. I'll let you know what I come up with.

Schnizz58
31-07-2015, 17:41
After further futzing with my spreadsheet, it looks like a scoop reduction of .1 and a knee of .4 will make that TX response curve pretty linear from a force of 2400 up. It looks exponential up to that point and then straightens out. Later on today I can post pics of my curves and the formuae in my spreadsheet if anybody is interested. Note that I haven't actually tried these values but I will do that too.

Edit: Correction....there's a goof in my spreadsheet so the above values are not correct. Let me try to fix it and I'll update this post.

Update: Ok, spreadsheet fixed. I played around with a few values and knee=0.48 and reduction=0.24 seem to look pretty linear. I will try these values in the game to see if I like them or not. I'll also try to find the settings that minimize the least square error between a perfectly linear output and the scoop processed output.

Another update: Looks like 0.54 and 0.24 are optimal from a linearity point of view. I tried driving with those settings and it's stronger overall and I think there's more feel under higher cornering loads. I'm going to keep these settings for a while and try them with different cars to see if I'm going to stick with them or not. Note that it was too strong for me so I lowered tire force to 52.

Other settings:
DDR 0.12
DRF 0.01 (two clicks to the right)
RAG 1.16
RAB 0.22
RAC 0.78

PlattanIMattan
31-07-2015, 22:29
After some fiddling with the excell files i finally got it to work. The 2 files in first post was broken and did not work at all. But after testing the file a few post down It finally worked.

Gonna be fun testing em out and see if it feels more linear.

poirqc
31-07-2015, 22:54
After some fiddling with the excell files i finally got it to work. The 2 files in first post was broken and did not work at all. But after testing the file a few post down It finally worked.

Gonna be fun testing em out and see if it feels more linear.

What do you mean by broken? If they're not working, i'll repost them.

Schnizz58
01-08-2015, 00:26
The orange line is the TX without scoop reduction, the blue line is after scoop reduction.
213951

inthebagbud
01-08-2015, 08:33
@schnizz

can you clarify some points for me to see if I am interpreting correctly

are you deriving the reduction point from where the torque and linear lines cross?

do you agree that on the torque line the knee position is around .46

how have you manipulated the figures to arrive at the corrected torque line

P1ckN1cker2406
01-08-2015, 12:08
Hi,

I did the Test with my Thrustmaster TX Wheel, but I dont know how to import it correctly into the sheet. Can you help me with this please?

Here are the csv files: 213972


I made the Test with the following Settings in the Thrustmaster Control Panel:

Overall FFB Strength: 65%
Constant: 100%
Periodic: 100%
Spring: 0%
Damping: 0%

unknwn
01-08-2015, 14:24
FFB gain in control panel impacts FFB linearity for TX/T300 wheels a lot. 60% gain is almost linear out of the box however max force will be lower compared to higher gain.
With 75% gain my wheel needs around 20-25 scoop reduction at around 55 knee( thought I am not really sure how knee should be interpreted.

poirqc
01-08-2015, 14:58
FFB gain in control panel impacts FFB linearity for TX/T300 wheels a lot. 60% gain is almost linear out of the box however max force will be lower compared to higher gain.
With 75% gain my wheel needs around 20-25 scoop reduction at around 55 knee( thought I am not really sure how knee should be interpreted.

I don't have numbers close but you're right about that. I saw a similar behavior with the G27. It's a nice thing to do the test a different force, then re sample the force that seems to give the better output.

At some point, i figured, for the G27, that it was probably better to let it at 100%

Cheers.

Schnizz58
01-08-2015, 15:41
@schnizz

can you clarify some points for me to see if I am interpreting correctly

are you deriving the reduction point from where the torque and linear lines cross?
I'm not sure I understand this question. I'm choosing a reduction and knee that yields the most linear output. I got it close just by trying different values and looking at the results. Once I was pretty close to a solution I tried all values that were "close" to the ones that looked good and the final values were the ones that minimized the difference from linear (in a least squares sense).


do you agree that on the torque line the knee position is around .46
Yes, that's in the right ballpark. 0.54 actually gave the best results.


how have you manipulated the figures to arrive at the corrected torque line
I just applied the scoop reduction algorithm (as I understand it) to the raw data. Look at the new graph below. I added two curves to the original graph. The light blue line just represents unity gain for reference. The dark red line is the scoop curve. It shows the gain applied to the raw data by the scoop algorithm. You can clearly see where the knee is. The knee is where the kink in the red line occurs and the reduction is the difference between the kink and the unity gain line. Note that this is based on the help text from the game, which says:

Scoops decrease low forces more and high forces less, thereby increasing the slope of force where some devices reduce the slope of force. The Scoop Knee is the force level where the slope changes from shallow to steep.

Scoop Reduction refers to the input force reduction below the knee. Above the knee, the force slope is increased such that at 1.0 input force, the output force is 1.0.

Hope that explains things better.

213979

unknwn
01-08-2015, 17:24
@Schnizz58
Could you share your worksheet file?

Schnizz58
01-08-2015, 17:50
Here you go. It's kind of in schnizz-ese so if there's anything you don't understand, just ask. The inputs are in cells s39 and s40 (shaded blue).

213990

b_akerlite
01-08-2015, 20:34
If we want to use this too do we paste in our values to columns A B C D and E? (from a step 2 force log check in wheelcheck?)
thanks!

Schnizz58
01-08-2015, 21:16
Correct. You actually only need columns A & C. Nobody ever explained to me what the other ones were for, so I didn't use them.

b_akerlite
01-08-2015, 22:35
Thanks unfortunately it doesnt work for me :(

https://docs.google.com/spreadsheets/d/1J0NQXN7hlVwxi8n40-ot5VXDC8h3Qorn_zAUDaZfCzs/edit?usp=sharing

I am almost there but my data is so different from yours?

Schnizz58
01-08-2015, 22:56
Looks like you're getting negative numbers out of the log2 analysis somehow. On the other hand, if you remove that big dead zone, yours looks pretty linear already.

b_akerlite
01-08-2015, 23:35
i know the 0.18 dead zone is unfortunate :( many hours played though!

I have set the 0.95 and 0.05 (which to be honest is probably as good as 0 and 0 !) and its the best of all the scoop settings i have tried.

not sure about the negative numbers - I have ran many log 2 force tests and I always get them...?

anyway thanks for looking at it and sorry for derailing the thread a bit!

poirqc
02-08-2015, 00:14
Well, in the original spreadsheet, the sheets named 100-X were samples, in the graph sheet, there's a column that averages thoses samples. Maybe Schnizz58 put them another way in his sheet.

Schnizz58
02-08-2015, 00:49
Well, in the original spreadsheet, the sheets named 100-X were samples, in the graph sheet, there's a column that averages thoses samples. Maybe Schnizz58 put them another way in his sheet.
That's entirely possible. Mine wasn't derived from yours; I started from scratch so I didn't do the averaging.

Schnizz58
02-08-2015, 15:30
After a few hours of testing, here's my issue with this "as linear as possible" approach. Big kerbs and minor off-course excursions feel pretty rough to me. For some examples, some of the bigger kerbs at Road America, such as the outside of the hairpin or at Brands Hatch, the outside of Stirling's. These spots threaten to jerk the wheel out of my hands if I run a little wide. So I went back to no scoop and it feels a little stiff at low to moderate forces. So what I decided to try was something in the middle. I used a knee of 0.4 and a reduction of 0.1. This takes advantage of the built-in signal compression of the TX but softens up the stiff feel at low steering angles. I also removed the relative torque adjust that I had been using. Now the only processing I'm using is deadzone removal and a little bit of scoop. No relative torque or soft clipping.

b_akerlite
02-08-2015, 15:31
Where could I go - who could I ask about the negative numbers I am getting in my log 2 test? Google doesn't throw up much on wheelcheck?

Schnizz58
02-08-2015, 15:35
Where could I go - who could I ask about the negative numbers I am getting in my log 2 test? Google doesn't throw up much on wheelcheck?
Your output is already pretty linear so I wouldn't do any scoop reduction at all. If you feel that kerbs and bumps are too rough, you may want to add some soft clipping. Start with SC half at 0.5 and SC full at 1.5 and go from there. I'd also add in some deadzone removal with a range of about 0.2 and whatever falloff value feels good to you.

b_akerlite
02-08-2015, 17:32
outside of the hairpin or at Brands Hatch
I swear I break my wrist every time I so much as sniff that curb - I think it is an exception though and not worth sacrificing the wheel settings to dial it out - Maybe sms could look at it cause it seems unusual -

yep I am loving my wheel again since I switched the scoop to 0.95 / 0.05 which is virtually off anyway.

Schnizz58
02-08-2015, 17:38
I swear I break my wrist every time I so much as sniff that curb - I think it is an exception though and not worth sacrificing the wheel settings to dial it out - Maybe sms could look at it cause it seems unusual -
It's a vicious kerb but it's like that in real life so I don't think SMS screwed up the model. The left-hander after Druids at Brands Hatch is the same way.

inthebagbud
04-08-2015, 06:31
FFB gain in control panel impacts FFB linearity for TX/T300 wheels a lot. 60% gain is almost linear out of the box however max force will be lower compared to higher gain.
.

I tested this and yes at 60% the line is pretty much linear, however it wasn't smooth as it had little steps at various points .

I actually liked it better at 75% with no knee or reduction

PTG Ty1er Ward
04-08-2015, 20:06
Thought I would throw this up for anyone interested. My guess is the physical weight of rim -

Stock Rim-
214244

GTE Rim-
214245

F1 Rim-
214246


TMCP all default (except F1 rim at 900DOR). We all pretty much new it anyway, but there's no difference between using a USB extender or not (I tested all 3 rims with and without). One more thing I want to try is to leave the default F1 Rim 360 DOR, set 360 in the test and see what happens.


Edit- for the sake of reference, I easily have 1000 hours of use on the TX base. I'm planning on using Schnizz's "theoretical/tinkering" worksheet next

PTG Ty1er Ward
06-08-2015, 16:34
Okay.....Who want's to know how deep the rabbit hole goes?

After dozens and dozens of tests (and 100s of CSV files) I don't even need the graph anymore, I can just look at the numbers LOL. Combined with in game FFB adjustments you do almost anything (within reason of your wheels capabilities). I learned so much I don't even know where to start, but I will say IMHO IMHO IMHO after all of this testing (on a TX) and also seeing other peoples graphs from Fanatec and Logitech users, then only possible way you could TRULY have "universal FFB settings" is to apply to much compression and gain that Forza FFB comes out the other side :)

I had on OK grasp of the settings prior to this, but its the complexity and interaction of the relationships that are now more clear to me. I am absolutely not an overnight expert- but I'm learning so I thought I'd post a few samples here for discussion if anyone wants. Since TMCP settings only apply to PC and I also play on Xbox (mostly actually) my personal in game FFB adjustments will be based on default (75- 100/100/100/100 which is pictured above) which is what I'm fairly certain the XBox uses. It's just easier for me to to have a commonality when bouncing back and forth between PC/Xbox. For PC only people here are some sample variations of TMCP settings and their linear response results.

214399
214400
214401
214402

If it were me personally only playing on PC, here's what I would use. It's the most linear I could get it without sacrificing a low deadzone:
214403

inthebagbud
06-08-2015, 17:11
PTG




From your post are you thinking that any changes to the TX control panel settings are ignored by the xbox, as I ran some tests with the gain setting and it did seem to change the wheel feel in game.


I did not explore too much as the having to unplug from xbox into pc and back again was a real pain and without an automatic way to actually test and compare results in controlled manner from the xbox it became to subjective to form a proper conclusion


Thus I applaud your time and effort to post your conclusions

PTG Ty1er Ward
06-08-2015, 17:34
PTG




From your post are you thinking that any changes to the TX control panel settings are ignored by the xbox, as I ran some tests with the gain setting and it did seem to change the wheel feel in game.


I did not explore too much as the having to unplug from xbox into pc and back again was a real pain and without an automatic way to actually test and compare results in controlled manner from the xbox it became to subjective to form a proper conclusion


Thus I applaud your time and effort to post your conclusions

Hi inthebagbud- Changes you make to the TMCP are to the driver for that PC and are not stored on the wheel, the Xbox is a different "PC/machine" and has it's own "driver" which we cannot access. A simple way to prove this to yourself is to set everything to Zero in the control panel and save, then connect it the Xbox you will still have FFB just like you always do on xbox but on PC the wheel would be dead :)

What you CAN change though using the mode button is the DOR and sensitivity (neither of which I recommend doing for PCars), but not the FFB gains

nissan4ever
06-08-2015, 21:35
PTG Ty1er Ward, I went ahead & tried your #'s for GTE rim for Scoop Knee & Reduction. Feels absolutely fantastic.

I'm curious, what is your RAG, RAB & RAC?

inthebagbud
06-08-2015, 22:24
Hi inthebagbud- Changes you make to the TMCP are to the driver for that PC and are not stored on the wheel, the Xbox is a different "PC/machine" and has it's own "driver" which we cannot access. A simple way to prove this to yourself is to set everything to Zero in the control panel and save, then connect it the Xbox you will still have FFB just like you always do on xbox but on PC the wheel would be dead :)

What you CAN change though using the mode button is the DOR and sensitivity (neither of which I recommend doing for PCars), but not the FFB gains

H'mmm going to have to do some testing on that - I have no knowledge on the xbox/tx control panel interaction , but if what you are saying is true it looks like I need to reevaluate

I am also curious as to how you came up with your scoop settings as everything seems to be based on input of 38, which I questioned earlier in thread as have others - how come this is fixed at 38 as a starting point

poirqc
06-08-2015, 22:37
That was the initial value i thought applied to the G27/G29. I didn't had the time to update the sheet to make it so someone would pick his knee and have a calculation from there.

If someone comes up with an updated sheet with something like this, i'll gladly update the first post.

Shortly, forget about the .38 knee value.

PTG Ty1er Ward
06-08-2015, 22:50
PTG Ty1er Ward, I went ahead & tried your #'s for GTE rim for Scoop Knee & Reduction. Feels absolutely fantastic.

I'm curious, what is your RAG, RAB & RAC?


I'm glad you found something you like but the math is Poirqc's (the OP) I just plugged in the numbers. Were you using higher numbers before? Now that I get it, I'm fairly certain at this point that any Scoop component application is the last thing the TX needs, especially on XBOX.

to quote from one of the early FFB setup guides-

"SCOOP-This is a new component for Topology 3, and is directly in response to some devices going flat in response at higher force levels. This is somewhat the opposite nonlinear tool as the soft clipper, but is shaped differently, So what scoop does is reduce lower forces more and high forces less, thereby increasing the slope of force where some devices reduce the slope of force."

This wheel would be the perfect scoop candidate:

214448



This is the TX (as far as Xbox is concerned) , you don't want to increase this slope, which is what the Scoop would do, you want it to level out quite a bit:

214450

PTG Ty1er Ward
06-08-2015, 23:09
H'mmm going to have to do some testing on that - I have no knowledge on the xbox/tx control panel interaction , but if what you are saying is true it looks like I need to reevaluate

Believe me my friend, that was first thing we all tried many moons ago to manipulate the FFB in Forza LOL :)

nissan4ever
07-08-2015, 01:09
This is the TX (as far as Xbox is concerned) , you don't want to increase this slope, which is what the Scoop would do, you want it to level out quite a bit

You might have just plugged in numbers or something. However, going from default

Scoop Knee 0.70
Scoop Reduction 0.15

Now it's this (GTE)

Scoop Knee 0.38
Scoop Reduction 0.21

I could feel a difference in a very positive way. I also run
FFB 76
Tire Force 61
Soft Clipping Half 0.60
Soft Clipping Full 1.60

Everything else is default, on the Force Feedback Calibration screen.

PTG Ty1er Ward
07-08-2015, 01:49
FFB feel is inarguably very personal in nature and whatever you or anyone likes is all that matters. That your changes to the scoop feel much better to you is actually in support of the math. If you look at the bottom numbers in the graph, you've moved the Knee from 70 to 38 which is just about the low force slope point anyway and you've change the force reduction below the knee (below 38) from 21% to a 15% reduction (only 7%). So really your almost running no scoop component anyway. You are spot on with the Clipper IMHO based on your main FFB and TF. That was what I was kind of thinking would be the right way to correct the curve. I was also thinking about fiddling with the damper setting which adds "weight" to the wheel. I couldn't identify any inherent drag with the TX and if you look at my test with each rim, the GTE was "less" rounded. I assumed due it being much heavier physically than the other two which gave me the damper idea.

Edit- adding a bit to hopefully help people understand the complexity of the system

Pre-2.0 there was a bug that if you changed the knee from away from 0.0 (the component is completely off at 0) you could never get it back to "off" because the change wouldn't save and you had to reset your main FFB settings. It would revert back to either 0.1 or 0.5 (can't remember where exactly) which is basically setting your knee there (0.1) meaning all the forces above your manually set knee are rescaled to 1.0 including those at 0.1 off of your deadzone which caused FFB craziness and violent shaking just off of center.

skoader
07-08-2015, 02:21
This is the TX (as far as Xbox is concerned) , you don't want to increase this slope, which is what the Scoop would do, you want it to level out quite a bit:

214450

Scoop really only increases the force slope above the knee. Below the knee it is reduced.

I'd be looking at a quite aggressive setting for that curve.

PTG Ty1er Ward
07-08-2015, 03:27
The purpose of the scoop isn't increasing forces above the knee (it rescales the forces below 1 to a force of 1). it's purpose is to reduce forces below your designated knee at your designated value. looking at the graph, with an input of 1 the TX produces an output much higher. Why would you compound the effect? (Honest question). Are you saying "aggressive" meaning a knee of 90-ish? I don't see how that would sense.

skoader
07-08-2015, 04:33
The purpose of the scoop isn't increasing forces above the knee (it rescales the forces below 1 to a force of 1). it's purpose is to reduce forces below your designated knee at your designated value.

That's not quite right. It rescales them so that an input force of 1 is output as 1.

Using data from inthebagbud's csv files back on page 2, I generated the following plot. ( Hope you don't mind :) )

214462

Scoop Knee here is 0.85 and Scoop Reduction is 0.37.

PTG Ty1er Ward
07-08-2015, 05:39
I don't mind at all, I'm happy to learn :) I unfortunately said things backwards in my last post LOL.

But yes, it rescales forces above the the knee to 1. It reduces forces below the knee (in your settings that's 85) by the amount you set (in your case 38%). But your graph represents what I'm trying to avoid. The TX doesn't have the same large deadzone the that Logitech has so introducing one by using a scoop doesn't make any sense to me. In your graph deadzone starts at 20% input there by squeezing/compressing 100% information into an 80% range and you don't reach 1 to 1 input/output until almost a 60% force of that shortened scale. It might feel great on a Logitech, but I promise you it doesn't on a TX. Everything below 60% force is well below 1:1 and the scoop is/has reduced/compressed those forces significantly (37%). In fact, I've already created a similar output scale using 0 scoop and only adjusting the TX control panel settings and the FFB feels exactly like the graph says it will: very weak with little detail until you pass the mid-range. Sure, that can be fixed using RAG, but now your just muddy'ing the pure signal even further. The point I'm trying to make is to take advantage of your particular wheels strengths, and use as little "corrections" as possible and not go down an endless compression/gain/compression/gain cycle until you have Forza/gran Turismo FFB. That's really what these "settings" are, they are linear "corrections" for consumer wheels. The wheels don't do the same thing so applying the same intricate and calculated settings don't make sense....Just my 2 cents

In a lot of ways the FFB system is just like tuning- there are over a dozen types of understeer/oversteer causes, all off which effect each other and other aspects of the handling as well. You're suggesting that on your car lowering the front springs is what helps, I'm implying that on my car (which is a completely different car) that is not the answer and makes laptimes slower.

skoader
07-08-2015, 06:35
I didn't say these settings apply to all wheels. They don't. My wheel wouldn't work well with them at all. ;)

They're generated specifically from inthebagbud's data. I assume we agree that the goal of the scoop is to get as near to a linear response from our wheels as possible?

inthebagbud
07-08-2015, 10:26
I assume we agree that the goal of the scoop is to get as near to a linear response from our wheels as possible?

I am getting confused now as to the actual end game here, I did think it was to move to a linear response, but after reading the last posts I wasn't sure?

Is it that we are discussing various wheels on various platforms - it would be useful if we could in the post list wheel and platform we are applying our thoughts too.?

Could we clarify in relation to the TX on Xbox and referring to my post http://forum.projectcarsgame.com/showthread.php?35147-Linear-FFB-Test-Excel-sheet-Universal&p=1064460&viewfull=1#post1064460

If the end game is to move to a more linear response how is this achieved, ignoring spreadsheets and formulas for the minute, as some of us need to understand the concept

With the TX/Xbox graph below

Scoop Knee- is the output figure of 0.48 derived by judging on the graph where the knee starts and drawing vertical line down to the input axis - is this correct

If so how do you move it to become more linear?

Do we draw a horizontal line from the TX output where the knee is ( in this case for the sake of example) 0.48 and where this intersects the linear output in this case say 0.80, is where we want to be?

Does this point 80 gives us the hypothetical knee required?

If we apply this on the Xbox this then I presume will distort the TX output, too much, and we have to apply a reduction to the below knee forces to compensate for this or do we ?

If we do is this derived from 0.80 minus 0.48 being 0.32

214469

Sorry if this is rather basic but some of us need to understand the underlying principle before moving to spreadsheet and formulas

inthebagbud
07-08-2015, 10:43
That was the initial value i thought applied to the G27/G29. I didn't had the time to update the sheet to make it so someone would pick his knee and have a calculation from there.

If someone comes up with an updated sheet with something like this, i'll gladly update the first post.

Shortly, forget about the .38 knee value.

Am I correct in my assumption that the basic idea is you find your knee point in the column headed input and then use the information on the same row to deduce scoop reduction with and without drr

hence and using row 21 in this example

Scoop Knee = input row 21
Scoop Reduction with DRR = output row 21 minus lineal row 21
Scoop Reduction without DRR = output row 21 minus lineal (No Drr) row 21

all divided by hundred to give a decimal?

The other variable I thought was the actual lineal column which would change depending on the wheel

Back in my original post I assumed

lineal step -this is derived from cell L51/45 - is this because cell L51 is where output = 100 and is divided by 45 because there are 5 steps before the mean value rises above the value of 1 (50 steps minus 5 = 45)

I am not saying I can write an excel to replicate this as we are into "IF" functions but I am sure if we clarify how the sheet should work somebody with more excel knowledge than me could do this easily

PTG Ty1er Ward
07-08-2015, 14:32
I assume we agree that the goal of the scoop is to get as near to a linear response from our wheels as possible?

We agree that Scoop is a tool that can be used to get as near as possible. What we don't seem to agree on is how is works and when to use it.


I am getting confused now as to the actual end game here, I did think it was to move to a linear response, but after reading the last posts I wasn't sure?

Is it that we are discussing various wheels on various platforms - it would be useful if we could in the post list wheel and platform we are applying our thoughts too.?

Could we clarify in relation to the TX on Xbox and referring to my post

If the end game is to move to a more linear response how is this achieved, ignoring spreadsheets and formulas for the minute, as some of us need to understand the concept

All of my graphs and data are from the TX, I split my time between Xobx and PC. If I only played on PC I would use the TM Control Panel to manage the linearity (you can see in some sample graphs I posted how close I was able to eventually get by only adjusting Main and Constant) and I would already be done with all of this. But since TMCP changes do not apply on the Xbox and I want the same FFB on each game I'm continuing on and starting to work with Clippers and Damping. I'll try to be as clear and concise as I can here: If you have a TX wheel you should A) never use a scoop - or - B) use it as a last resort (its comes virtually dead last in the signal path/chain anyway). Using it as a starting point on the TX is wrong (probably with other wheels too). You are much better served using the Clipper and Gain tools for as close to linearity with the TX (probably other wheels too).
.
The graphs I (and others) have shown are the pure input/output with no corrections. If your in game settings include any Clipper or Relative Adjust then your linear graph would not nearly be the same, so if you are trying to use a pure input/ouput graph to implement a scoop your are not even looking at the right data. The data absolutely NAILS the deadzone if you match it with your FFB setting. 1,000,000 times better

If you have 0.0 Relative Gain and 0.0 Clipper and a 100 FFB setting then YES, you could technically use the data for identifying how to implement a scoop. BUT the graph clearly shows scoop is the wrong tool to use for the TX wheel base output curve. To put it into tuning terms- Your adding oversteer to an oversteer car trying to fix oversteer.

Even if its used dead last as a final adjustment, Scoop may or may not be of use to the TX because in all of my testing I was never able to create an output that fit the profile the Scoop tool was actually intended for, but I can't see the data with Clippers and other corrections applied so like I said, it may be of some use at the end.

skoader
07-08-2015, 15:12
I give up. Carry on...

Schnizz58
07-08-2015, 16:20
That's not quite right. It rescales them so that an input force of 1 is output as 1.

Using data from inthebagbud's csv files back on page 2, I generated the following plot. ( Hope you don't mind :) )

214462

Scoop Knee here is 0.85 and Scoop Reduction is 0.37.
You're on the right track here but your linear force and linear force response are scaled incorrectly (I think). The scoop curve is correct. At maximum force, the TX only gives you about 75-80% of what you ask for. So you need to scale that to 0.75-0.8 instead of 1.0. Below is the graph I created from my TX data to illustrate what I'm talking about.

214509

Also you asked about the ultimate goal of this process. I think everybody's goal is a little different due to the personal nature of FFB. Some might want as linear an output as possible. I tried that and found that there were things I didn't like about it (very strong forces were too strong) and so I opted for something in between a linear response and the flattened curve you get without any scoop at all. Scoop is just a tool in the toolkit and you can use it to shape the response curve in certain ways. So to me the goal is to understand how scoop works and how to use it to shape the response curve to your liking.

PTG Ty1er Ward
07-08-2015, 17:02
You're on the right track here but your linear force and linear force response are scaled incorrectly (I think). The scoop curve is correct. At maximum force, the TX only gives you about 75-80% of what you ask for. So you need to scale that to 0.75-0.8 instead of 1.0. Below is the graph I created from my TX data to illustrate what I'm talking about.

Also you asked about the ultimate goal of this process. I think everybody's goal is a little different due to the personal nature of FFB. Some might want as linear an output as possible. I tried that and found that there were things I didn't like about it (very strong forces were too strong) and so I opted for something in between a linear response and the flattened curve you get without any scoop at all. Scoop is just a tool in the toolkit and you can use it to shape the response curve in certain ways. So to me the goal is to understand how scoop works and how to use it to shape the response curve to your liking.


After a few hours of testing, here's my issue with this "as linear as possible" approach. Big kerbs and minor off-course excursions feel pretty rough to me. For some examples, some of the bigger kerbs at Road America, such as the outside of the hairpin or at Brands Hatch, the outside of Stirling's. These spots threaten to jerk the wheel out of my hands if I run a little wide. So I went back to no scoop and it feels a little stiff at low to moderate forces.

This is what I've been trying to explain (very POORLY apparently). Scoop Knee – the force level where the force changes from shallow to steep. The TX output is STEEP TO SHALLOW, that's the opposite. I interpret this to mean in the case of the TX your are Steepening an already Steep Slope. It also accounts for your descriptions of the effects above. I'm suggesting that-

Above the Knee point, all forces that are below a 1.0 input are scaled to at 1.0 input- so a kerb that would normally come through as a 60% force is scaled to a 100% force (assuming your knee point is set below 60)

I know that the guide it says "Above the knee, the force slope is increased such that at 1.0 input force, the output force is 1.0" meaning no change, but it also says-

This is a new component for Topology 3, and is directly in response to some devices going flat in response at higher force levels. This is somewhat the opposite nonlinear tool as the soft clipper, but is shaped differently, to better fit the nature of devices. So what scoop does is reduce lower forces more and high forces less, thereby increasing the slope of force where some devices reduce the slope of force."

"Scoops increase the slope of force for some devices that reduce the slope of force". - This doesn't describe the TX, it describes a Wheel that outputs like this:

214518


Can a DEV who was involved if the FFB design shed any light here? (pretty please :))

Schnizz58
07-08-2015, 17:08
I know that the guide it says "Above the knee, the force slope is increased such that at 1.0 input force, the output force is 1.0" meaning no change, but it also says-

This is a new component for Topology 3, and is directly in response to some devices going flat in response at higher force levels. This is somewhat the opposite nonlinear tool as the soft clipper, but is shaped differently, to better fit the nature of devices. So what scoop does is reduce lower forces more and high forces less, thereby increasing the slope of force where some devices reduce the slope of force."
You seem to imply here that these two statements are at odds with one another but I find them to be perfectly consistent. Am I misinterpreting your statement?


"Scoops increase the slope of force for some devices that reduce the slope of force". - This doesn't describe the TX
On the contrary, that describes the TX perfectly. The curve flattens out around halfway up as shown in the graphs of TX response curves in this thread (mine and others).

PTG Ty1er Ward
07-08-2015, 17:54
You seem to imply here that these two statements are at odds with one another but I find them to be perfectly consistent. Am I misinterpreting your statement?


On the contrary, that describes the TX perfectly. The curve flattens out around halfway up as shown in the graphs of TX response curves in this thread (mine and others).

No, you're not misinterpreting, I believe those to statements say the same thing, but I'm suggesting that forces above the Knee and below 1.0 are not 1:1 and that all forces below 1.0 scaled to 1.0 (above the knee)

This is that part that everyone has me feeling like I need to check into a mental hospital LOL.

Yes the TX curve flattens out around halfway, but the curve it is STEEP TO SHALLOW Scoop is a "one way" tool is intended to correct to a curve that is shaped the opposite of the TX (the opposite of the TX curve being shallow to steep)

Scoop Knee – the force level where the force changes from shallow to steep. The TX curve is STEEP TO SHALLOW:D

"This is a new component for Topology 3, and is directly in response to some devices going flat in response at higher force levels. SCOOP IS THE SOMEWHAT OPPOSITE NON LINEAR TOOL AS THE SOFT CLIPPER":D:D:D

I am saying Clippers are the appropriate correction tool for the TX (striving for linearity)

Schnizz58
07-08-2015, 18:59
Yes the TX curve flattens out around halfway, but the curve it is STEEP TO SHALLOW Scoop is a "one way" tool is intended to correct to a curve that is shaped the opposite of the TX (the opposite of the TX curve being shallow to steep)
No, it's meant to correct exactly that type of response. TX goes steep to shallow, scoop goes shallow to steep, put them together and you get a straight line (approximately).


I am saying Clippers are the appropriate correction tool for the TX (striving for linearity)
Sorry you have it backwards. The TX curve is essentially like soft clipping already so adding more soft clipping just makes the curve that much flatter.

PTG Ty1er Ward
07-08-2015, 19:51
No, it's meant to correct exactly that type of response. TX goes steep to shallow, scoop goes shallow to steep, put them together and you get a straight line (approximately).


Sorry you have it backwards. The TX curve is essentially like soft clipping already so adding more soft clipping just makes the curve that much flatter.

hmmmmm, I'll be DAMNED LOL Thanks Schnizz!!!!!!!!!!!!!!!!!!!! (and Skoader gave up on me :hopelessness: LOL)

So it's a 2 part process then, that's makes sense. A 50% Scoop Reduction is actually applied to across the entire range (the Blue line becomes the Purple Line), NOT just below the Knee. Then "Above the knee, the force slope is increased such that at 1.0 input force, the output force is 1.0" The Red Circle/Red Line. Like this:

214533

Schnizz58
07-08-2015, 21:04
Sorta. Scroll up and look at my graph in post #99. The red line shows the scoop function (with knee = 0.54 and reduction = 0.24). See how it has a kink in it? The slope of the line is shallow up to the kink and then gets steeper so that an input force of 1.0 produces an output force of 1.0. (My spreadsheet actually calculates the slopes of those lines.) On that same graph the orange line is the force response of the TX wheel. The idea is that knowing how the TX response flattens out, you make the force from the game nonlinear in such as way as to make the force you feel seem linear. So that means that the scoop function needs a kink in it right about where the response curve flattens out but in the opposite direction to compensate. Now look at the blue-green curve labeled Corrected Torque. This should be pretty close to the perceived force you feel after the scoop reduction processing. I played around with values of knee and reduction to make that as linear as I could and those values (0.54 & 0.24) were the ones that were most linear.

However when I tried those values in the game, I didn't like the result. To get the same amount of feel at low forces that I had before, I needed to increase tire force. (Because the slope of the corrected curve is shallower than the slope of the uncorrected curve in that region.) But when I increased the tire force, large forces felt too large. So it turns out that I actually liked a little bit of flattening (which is basically compression or soft clipping). I picked values that gave me something in between the raw response and a perfectly linear response. Now I think I have the best of both worlds. I have good detail for low to moderate forces without too strong of a force when I hit a big kerb (for example the kerb on the outside of Stirling's at Brands Hatch). The values I have settled on for now are 0.42 and 0.18 I think (not in front of the game at the moment but they're somewhere in that vicinity).

poirqc
07-08-2015, 22:09
Am I correct in my assumption that the basic idea is you find your knee point in the column headed input and then use the information on the same row to deduce scoop reduction with and without drr

hence and using row 21 in this example

Scoop Knee = input row 21
Scoop Reduction with DRR = output row 21 minus lineal row 21
Scoop Reduction without DRR = output row 21 minus lineal (No Drr) row 21

all divided by hundred to give a decimal?

The other variable I thought was the actual lineal column which would change depending on the wheel

Back in my original post I assumed

lineal step -this is derived from cell L51/45 - is this because cell L51 is where output = 100 and is divided by 45 because there are 5 steps before the mean value rises above the value of 1 (50 steps minus 5 = 45)

I am not saying I can write an excel to replicate this as we are into "IF" functions but I am sure if we clarify how the sheet should work somebody with more excel knowledge than me could do this easily

I'll try to answer you when my brain kicks in. The more i read about the subject, the more i get lost! :D

Schnizz58
07-08-2015, 22:26
Am I correct in my assumption that the basic idea is you find your knee point in the column headed input and then use the information on the same row to deduce scoop reduction with and without drr
A good starting point would be to pick the knee where the delta between the response curve and the linear output is greatest. Then pick a reduction value that will pull the response curve down to the value of the linear output at the point of the knee. Then you can play around with values adjacent to those to see if they improve linearity. Nothing I've read defines exactly how much more force is added in for deadzone removal so I have no way to model that. You can change the range that the deadzone removal works on and how fast it falls off but there is no parameter that specifies the additional gain applied in the deadzone removal range. Also note that deadzone removal occurs after scoop in the processing chain.


I am not saying I can write an excel to replicate this as we are into "IF" functions but I am sure if we clarify how the sheet should work somebody with more excel knowledge than me could do this easily
That's basically what I did in my spreadsheet.

inthebagbud
08-08-2015, 05:09
I give up. Carry on...

Skoader you can come back now ;)

inthebagbud
08-08-2015, 05:31
You're on the right track here but your linear force and linear force response are scaled incorrectly (I think). The scoop curve is correct. At maximum force, the TX only gives you about 75-80% of what you ask for. So you need to scale that to 0.75-0.8 instead of 1.0. Below is the graph I created from my TX data to illustrate what I'm talking about.

Schnizz58 just so i understand on what basis do you believe that the tx output is only 75-80% and not 100%

Schnizz58
08-08-2015, 13:48
Look at the data. The input is 10000 and the output is 7500.

Fre.Mo
08-08-2015, 16:45
How can we be sure that the wheelcheck is representative of project cars output too?

Schnizz58
08-08-2015, 17:06
How can we be sure that the wheelcheck is representative of project cars output too?
What do you mean by representative of project cars output? It's just a number that the game sends to the wheel. The wheel doesn't know whether it came from a game or from a test program.

skoader
08-08-2015, 22:39
You're on the right track here but your linear force and linear force response are scaled incorrectly (I think). The scoop curve is correct. At maximum force, the TX only gives you about 75-80% of what you ask for. So you need to scale that to 0.75-0.8 instead of 1.0. Below is the graph I created from my TX data to illustrate what I'm talking about.

214509



I had a look at the excel file you posted and see you're using the endX data to arrive at that 75% figure. EndX is just a measure of the wheels position at the end of each step - In each step a constant force is applied for 300ms.
100% force doesn't mean your wheel should rotate all the way to lock. It's the relative values that are important.

Also, your corrected torque calculation isn't right -
At 50% input force the wheel responds with deltaX value of 65%.
With your knee and reduction settings, an input force of around 67% is correctly 'scooped' to 50% and should produce the same response as above, but you have a much lower value there.

Schnizz58
08-08-2015, 22:52
I had a look at the excel file you posted and see you're using the endX data to arrive at that 75% figure. EndX is just a measure of the wheels position at the end of each step - In each step a constant force is applied for 300ms.
100% force doesn't mean your wheel should rotate all the way to lock. It's the relative values that are important.
Thanks. What values should I be using?


Also, your corrected torque calculation isn't right -
At 50% input force the wheel responds with deltaX value of 65%.
With your knee and reduction settings, an input force of around 67% is correctly 'scooped' to 50% and should produce the same response as above, but you have a much lower value there.
Can you elaborate? I'm not following. Also, do you know what the 5 different outputs represent? I tried to get someone to explain them to me earlier in this thread but nobody would. So I Just made the assumption that endX is what I should be using.

Schnizz58
08-08-2015, 22:56
Pre-2.0 there was a bug that if you changed the knee from away from 0.0 (the component is completely off at 0) you could never get it back to "off" because the change wouldn't save and you had to reset your main FFB settings. It would revert back to either 0.1 or 0.5 (can't remember where exactly) which is basically setting your knee there (0.1) meaning all the forces above your manually set knee are rescaled to 1.0 including those at 0.1 off of your deadzone which caused FFB craziness and violent shaking just off of center.
That was soft clipping not scoop.

skoader
09-08-2015, 23:10
Can you elaborate? I'm not following. Also, do you know what the 5 different outputs represent? I tried to get someone to explain them to me earlier in this thread but nobody would. So I Just made the assumption that endX is what I should be using.

startX = Wheel position before the step.
endX = Wheel position after the step.
deltaX = endX - startX.
deltaXDeg = deltaX converted to degrees of rotation. (You must have 'Wheel angle' set correctly in wheelcheck. Default is 900)

So the last 2 are the only useful entries for our purposes and are essentially the same thing just scaled differently.
I use deltaXDeg but rescale it so that the maximum reported rotation is 1, which is fine for analyzing a single curve. Not if you want to compare results with different driver settings for example.

Using the excel file you posted as reference and sticking with endX values as that's what your curves are built on -

Your normalized scoop forces (column G) match mine exactly.
We know from the linear force test that the wheel responds to 50% force by rotating to a little under 66% (C27).
If we look up your scoop force at 50% (Right in between G35-G36) the wheel response at that point for your 'corrected torque' curve should be the same 66%. Your calculated response however is just 53%. (Between I35-36)

I often find it difficult to express myself clearly in text. I hope that's better.

Schnizz58
10-08-2015, 00:16
Great explanation, thanks very much. Wasn't your fault at all, I just needed more detail. I will digest this information and redo the spreadsheet.

poirqc
10-08-2015, 02:35
Great explanation, thanks very much. Wasn't your fault at all, I just needed more detail. I will digest this information and redo the spreadsheet.

Id like to see that also because i understand the concept of the Knee. I still have trouble understanding how the corrected value is applied.

If we could get a calculation about the Deadzone Removals and how the 2 gets together!

Keep up the good work guys!

Lars Rosenquist
10-08-2015, 10:30
How can we be sure that the wheelcheck is representative of project cars output too?

As long as you make sure FFB strength is at the same level in both, you're good. E.g. if you use 100% FFB level in pCARS, make sure it's also tested at 100% FFB strength in Wheelcheck (e.g. for Logitech this means 100% effects strength in the profiler).

inthebagbud
10-08-2015, 11:41
As long as you make sure FFB strength is at the same level in both, you're good. E.g. if you use 100% FFB level in pCARS, make sure it's also tested at 100% FFB strength in Wheelcheck (e.g. for Logitech this means 100% effects strength in the profiler).

Good point from Fre.Mo question as I am sure the TMCP defaults to 75%

Lars Rosenquist
10-08-2015, 12:58
Right now FFB defaults to 100 in Project CARS, but it used to be 85. E.g. for the G27, at 100% FFB, FFB Deadzone is 16%, but at 85% this becomes 35-ish, so it's definitely something to take into account.

jhonatas
10-08-2015, 17:40
You're driving me crazy! I've been following this whole topic, I use the wheel T300 I believe that is similar to the TX, I really do not know what to do.
First I used TF 80% Sccop Knee 75, reduction 15, today changed and took Sccop Knee - 0 and Reduction - 0, but the wheel got heavier, then the TF to 54%. I have the impression that I feel more low strength, but not even sure of that.

inthebagbud
10-08-2015, 17:57
You're driving me crazy! I've been following this whole topic, I use the wheel T300 I believe that is similar to the TX, I really do not know what to do.
First I used TF 80% Sccop Knee 75, reduction 15, today changed and took Sccop Knee - 0 and Reduction - 0, but the wheel got heavier, then the TF to 54%. I have the impression that I feel more low strength, but not even sure of that.

At moment this is a work in progress and subject too a lot of discussion and learning from all the contributors. As new things are learnt and discussed then the current thinking will change so best just too follow the thread for moment.

Schnizz58
10-08-2015, 23:56
skoader I changed my spreadsheet to use deltaX but it's pretty much the same as endX anyway. Other than that, I don't see where I made any mistakes. One difference I noticed is that you're normalizing the response to 1.0 and I'm not. In other words for a full output of 10000, you're expecting a response of 10000 but we know that isn't the case. I don't know if that accounts for the difference or not.

My equation for corrected torque is
output torque = deltaX * scoop / (input torque)

Is that incorrect?

inthebagbud
11-08-2015, 04:48
skoader I changed my spreadsheet to use deltaX but it's pretty much the same as endX anyway. Other than that, I don't see where I made any mistakes. One difference I noticed is that you're normalizing the response to 1.0 and I'm not. In other words for a full output of 10000, you're expecting a response of 10000 but we know that isn't the case. I don't know if that accounts for the difference or not

This is the bit that is confusing me as poirgc spreadsheet normalises to 1 to 1 by manipulating the deltaxdeg figures to provide the output figures. I get the logic of what is going on but struggle with understanding the spreadsheet manipulation so my thanks to Schnizz58 and skoader for carrying on the discussion.

For my contribution if you need any tests running in game just let me know

SeeNoWeevil
11-08-2015, 11:45
What does the FFB strength setting in your wheel profiler actually change? If the game sends 100% to my wheel, what value in my Thrustmaster profiler would I set if I wanted that to equate to 100% FFB motor output? Or does the setting actually change the linearity of the wheel's output and 100% in is always 100% out?

How does wheelcheck.exe actually know how much force your wheel is outputting for a given input?

Apologies if it's already been mentioned, but what settings am I changing in pCars based on what I see in my output graph to make the response more linear? Wouldn't this then invalidate any *other* settings I've applied, for example bManic's posted here.

It would be really nice if someone could edit the initial post with some sort of guide on how to use the data and what the thread is about.


Thanks

inthebagbud
11-08-2015, 12:07
What does the FFB strength setting in your wheel profiler actually change? Or does the setting actually change the linearity of the wheel's output and 100% in is always 100% out?

As I understand in the Xbox the TX control panel settings are ignored by the game - how the wheel then interprets this information is the grey area


As long as you make sure FFB strength is at the same level in both, you're good. E.g. if you use 100% FFB level in pCARS, make sure it's also tested at 100% FFB strength in Wheelcheck (e.g. for Logitech this means 100% effects strength in the profiler).

Lars was pointing out that if we are using wheelchecker, to make sure that the gain setting in the control panel matched that of the game or the information we are basing any assumptions on would be incorrect

poirqc
11-08-2015, 12:11
This is the bit that is confusing me as poirgc spreadsheet normalises to 1 to 1 by manipulating the deltaxdeg figures to provide the output figures. I get the logic of what is going on but struggle with understanding the spreadsheet manipulation so my thanks to Schnizz58 and skoader for carrying on the discussion.

For my contribution if you need any tests running in game just let me know

I changed the range of the value after a discussion with skoader in the G29 thread. The ranges in pCars are 0 to 1 so i thought it would be better to show the data as such. Everything would be aligned.

Schnizz58
11-08-2015, 12:57
The maximum output of the test is 10000 units of something, what I don't know but it doesn't matter. With that output I get about 7500 units of response from the wheel. So I don't think it's correct to scale that 1:1. I didn't normalize my spreadsheet; I just left it in the raw units.

inthebagbud
11-08-2015, 14:32
The maximum output of the test is 10000 units of something, what I don't know but it doesn't matter. With that output I get about 7500 units of response from the wheel. So I don't think it's correct to scale that 1:1. I didn't normalize my spreadsheet; I just left it in the raw units.

Out of curiosity what is the gain setting in your TX Control Panel, if it is at default 75% could this be the answer that the wheel when used in wheelchecker is only giving 75% response

Just a thought to eliminate any other factors at work

Schnizz58
11-08-2015, 14:43
Out of curiosity what is the gain setting in your TX Control Panel, if it is at default 75% could this be the answer that the wheel when used in wheelchecker is only giving 75% response

Just a thought to eliminate any other factors at work
I've never changed it because the control panel doesn't run on my computer (much like wheelchecker). Is the default 75? That certainly would explain it.

inthebagbud
11-08-2015, 17:20
I've never changed it because the control panel doesn't run on my computer (much like wheelchecker). Is the default 75? That certainly would explain it.

I will run the wheelchecker with the TX panel at 100% and upload the data, if you want so you can run it through your sheet, but it may be a couple of days as got to to take my pit lady out tonight.

Its a hard life PCars, beer and a pit lady!

PTG Ty1er Ward
11-08-2015, 18:15
I will run the wheelchecker with the TX panel at 100% and upload the data, if you want so you can run it through your sheet, but it may be a couple of days as got to to take my pit lady out tonight.

Its a hard life PCars, beer and a pit lady!

I did that (and dozens of other variations LOL) last week. This particular test is 1 set (not an average of 4) but I haven't ever seen something more than negligibly different. I figured out how to get the test to run 4 consecutive times afterwards anyway. Here you go:

214862

Increasing the main to 100 nearly halved the deadzone.

Which brings up another interesting point- I thought of this last week but your just reminded me. In TMCP you have Main FFB, then 4 individual settings; Constant, Periodic, Spring, Damper. Main is obviously the master "volume", but Constant seemed to have roughly the same effect. A Zero Constant setting output Zero FFB (even with Main at 100). I kind of equated to the same relationship in PCars as Main and Tire Force (could be wrong obviously since I was so good at understanding Scoop in the beginning :rolleyes:). The physical weight of the wheel has an effect as well, but not much, maybe a few percent.

poirqc
11-08-2015, 19:16
I did that (and dozens of other variations LOL) last week. This particular test is 1 set (not an average of 4) but I haven't ever seen something more than negligibly different. I figured out how to get the test to run 4 consecutive times afterwards anyway. Here you go:

214862

Increasing the main to 100 nearly halved the deadzone.

Which brings up another interesting point- I thought of this last week but your just reminded me. In TMCP you have Main FFB, then 4 individual settings; Constant, Periodic, Spring, Damper. Main is obviously the master "volume", but Constant seemed to have roughly the same effect. A Zero Constant setting output Zero FFB (even with Main at 100). I kind of equated to the same relationship in PCars as Main and Tire Force (could be wrong obviously since I was so good at understanding Scoop in the beginning :rolleyes:). The physical weight of the wheel has an effect as well, but not much, maybe a few percent.

Increasing or decreasing the "master volume" on the device control panel of a wheel will move the FFB deadzone accordingly. I observed that also on my G27.

inthebagbud
11-08-2015, 19:32
But according to previous posts these settings will have no effect on the xbox?

Presume you are talking about in the pc version.

SeeNoWeevil
11-08-2015, 20:22
http://i.imgur.com/XhJqjwV.png

I've read the thread from start to finish and am still none the wiser how to take this information from my T300 and get the right settings for pCars to give me a more linear response. Anyone help out?

Schnizz58
11-08-2015, 20:45
http://i.imgur.com/XhJqjwV.png

I've read the thread from start to finish and am still none the wiser how to take this information from my T300 and get the right settings for pCars to give me a more linear response. Anyone help out?
If that's the response curve of your T300, just eyeballing it I'd start with a knee around 0.6-0.65 and a reduction of about 0.2-0.25. Also a deadzone of about 0.10.

PTG Ty1er Ward
11-08-2015, 21:05
But according to previous posts these settings will have no effect on the xbox?

Presume you are talking about in the pc version.

Yes, on PC. I'm just throwing out the idea that the TMCP Main and Constant settings could possibly correlate to the PCars Main and Tire Force settings. It seemed to me that the Main and Constant settings have the biggest direct impact on the "out of the box" linearity for the TX based on my testing. If the settings are found to generally be mirrors of one another, than that does indeed help the Xbox users (myself included).

SeeNoWeevil
11-08-2015, 21:11
If that's the response curve of your T300, just eyeballing it I'd start with a knee around 0.6-0.65 and a reduction of about 0.2-0.25. Also a deadzone of about 0.10.

Thanks for that. Can I ask how you came to those numbers so I can change my profiler strength and redo them? Is one Thrustmaster profiler strength value a better starting point in terms of being more linear? I'm a bit confused why wheel manufacters don't smooth things out at a driver level.

PTG Ty1er Ward
11-08-2015, 21:12
Increasing or decreasing the "master volume" on the device control panel of a wheel will move the FFB deadzone accordingly. I observed that also on my G27.

Is there a significant downside to removing a larger deadzone?

It would seem to me that it would be better to work with more a "pure" signal, even if its a bit "weaker", rather than compressing/scooping/correcting a stronger one (or I'm on the completely wrong side of the fence again lol)

PTG Ty1er Ward
11-08-2015, 21:19
Is one Thrustmaster profiler strength value a better starting point in terms of being more linear?

I'm basically asking this exact question in the post above. Depending on the answer we get I can tell you the "profiler" TMCP settings that make the TX as linear as I could get it. They may work well for the T300 too

Schnizz58
11-08-2015, 21:23
Thanks for that. Can I ask how you came to those numbers so I can change my profiler strength and redo them? Is one Thrustmaster profiler strength value a better starting point in terms of being more linear? I'm a bit confused why wheel manufacturers don't smooth things out at a driver level.
The knee goes near the point where the delta between the response curve and the linear line is a maximum. The reduction is just the amount you want to pull down the curve to try to match the line. Like I said, it's just an eyeball guess. You will want to try other values near those to see if you like them better. You may also want to leave some non-linearity in (I did). In fact, I might just go with no scoop at all.

Sorry about your profiler question. I don't know anything about TM profilers.

I wouldn't want the manufacturers to smooth anything out. It's up to the game to do that.

SeeNoWeevil
11-08-2015, 21:51
The knee goes near the point where the delta between the response curve and the linear line is a maximum. The reduction is just the amount you want to pull down the curve to try to match the line. Like I said, it's just an eyeball guess. You will want to try other values near those to see if you like them better. You may also want to leave some non-linearity in (I did). In fact, I might just go with no scoop at all.

Sorry about your profiler question. I don't know anything about TM profilers.

I wouldn't want the manufacturers to smooth anything out. It's up to the game to do that.
Thanks. Smooth out was probably the wrong word. Why don't they achieve a linear output is more what I was suggesting. Or is it like audio equipment where a 'boosted' output is more pleasing to a lot of users?

poirqc
11-08-2015, 22:18
Is there a significant downside to removing a larger deadzone?

It would seem to me that it would be better to work with more a "pure" signal, even if its a bit "weaker", rather than compressing/scooping/correcting a stronger one (or I'm on the completely wrong side of the fence again lol)

Well, conceptualy, since i'm still not sure about numbers, is that deadzone removal should be set precisely. With Deadzone removal, the only goal, imho, should be to only remove the values generated by the game where the wheel can't respond to that.

If you don't remove enough, the game will still generate values the wheel can't output.
If you remove too much, the wheel won't output some fine torques changes you could feel.

P.S.: You owe me a glass of stout! I spilled mine when i tried to spell check! :D

Schnizz58
11-08-2015, 22:27
Thanks. Smooth out was probably the wrong word. Why don't they achieve a linear output is more what I was suggesting. Or is it like audio equipment where a 'boosted' output is more pleasing to a lot of users?
It could just be a limitation of the equipment. To get a linear output might take more expensive components. I've noticed that Fanatec wheels look pretty linear but they're 2-3x the cost of a TX. Or it could be that it's tuned that way on purpose. When I tried for a linear output, I didn't really like the feel and ended up with something that was not as flat as the normal TX response but not truly linear either. If I had a linear wheel like a Fanatec, I'd probably have to add some soft clipping to flatten out the upper part of the curve.

PTG Ty1er Ward
11-08-2015, 22:52
Well, conceptualy, since i'm still not sure about numbers, is that deadzone removal should be set precisely. With Deadzone removal, the only goal, imho, should be to only remove the values generated by the game where the wheel can't respond to that.

If you don't remove enough, the game will still generate values the wheel can't output.
If you remove too much, the wheel won't output some fine torques changes you could feel.

P.S.: You owe me a glass of stout! I spilled mine when i tried to spell check! :D

You don't want Stout that comes from Utah, believe me :)


It could just be a limitation of the equipment. To get a linear output might take more expensive components. I've noticed that Fanatec wheels look pretty linear but they're 2-3x the cost of a TX. Or it could be that it's tuned that way on purpose. When I tried for a linear output, I didn't really like the feel and ended up with something that was not as flat as the normal TX response but not truly linear either. If I had a linear wheel like a Fanatec, I'd probably have to add some soft clipping to flatten out the upper part of the curve.

After spending some time in the iRacing Forum I'm not 100% positive we're using the Step2 data the right way. What we've been doing is running 4 separate tests of the same FFB setting (like 75 default or whatever you have) then averaging those 4 values.

It seems what everyone is doing there is using data from multiple FFB points (ie 25%/50%/75%100%) then averaging THOSE values or using them in a different mathematical manner possibly. I've even seen some that even have data for every 5% step.

Here's an example of someone's T500 step log data

214887

Schnizz58
11-08-2015, 23:05
You don't want Stout that comes from Utah, believe me :)
I don't know about that. I've had some pretty decent beers in SLC.


After spending some time in the iRacing Forum I'm not 100% positive we're using the Step2 data the right way. What we've been doing is running 4 separate tests of the same FFB setting (like 75 default or whatever you have) then averaging those 4 values.
Wouldn't surprise me that we've been doing it wrong. I've only been doing a single run but I want to change my FFB setting to 100 and rerun it. Also I wonder if any of the other tests in that program are useful.

skoader
11-08-2015, 23:09
skoader I changed my spreadsheet to use deltaX but it's pretty much the same as endX anyway. Other than that, I don't see where I made any mistakes. One difference I noticed is that you're normalizing the response to 1.0 and I'm not. In other words for a full output of 10000, you're expecting a response of 10000 but we know that isn't the case. I don't know if that accounts for the difference or not.

No it doesn't, but I think I understand the source of confusion. I'll post details later today, we really need to be on the same page here before moving forward.



My equation for corrected torque is
output torque = deltaX * scoop / (input torque)

Is that incorrect?

Yeah that's incorrect. Again, details to come...

Schnizz58
11-08-2015, 23:18
What a tease... :)

poirqc
11-08-2015, 23:19
You don't want Stout that comes from Utah, believe me :)



After spending some time in the iRacing Forum I'm not 100% positive we're using the Step2 data the right way. What we've been doing is running 4 separate tests of the same FFB setting (like 75 default or whatever you have) then averaging those 4 values.

It seems what everyone is doing there is using data from multiple FFB points (ie 25%/50%/75%100%) then averaging THOSE values or using them in a different mathematical manner possibly. I've even seen some that even have data for every 5% step.

Here's an example of someone's T500 step log data

214887

Before starting this thread, i did some sheets. One of them was testing my G27 with various force set into the logitech control panel. From there i saw the different ways the wheel behaved depending on the strenght of the Logitech control panel. At some point i figured that changing the strenght, for the G27, would only offset the deadzone and the response curve. The shape would remain the same. After that, i multi sampled a couple force and saw that the response curve was often similar. I then figured most people would have it at 100%, so if i posted some FFB setup, it could work.

I think it's logicaly sound to align tests and play values. You set the control panel to a given strenght, you test it with wheelcheck with that value, you then test pCars with that same value.

When you pick a value, you then generate couple samples to discard electrical or mecanical glitches.

That was, in essence, my train of thoughts. All thoses test are spread in couple threads on this forum.

Cheers!

poirqc
11-08-2015, 23:22
No it doesn't, but I think I understand the source of confusion. I'll post details later today, we really need to be on the same page here before moving forward.



Yeah that's incorrect. Again, details to come...

I agree

An sms hint would really help us :D I don't think my brain can process more theories or graph anymore! :D

skoader
11-08-2015, 23:30
I agree

An sms hint would really help us :D I don't think my brain can process more theories or graph anymore! :D

Be warned, graphs incoming. lol

Schnizz and I largely agree on how the scoop works. It's graphing the wheels response to those scooped forces where we differ at the moment. Not really something SMS can help with.

PTG Ty1er Ward
12-08-2015, 00:18
I think what has turned me upside down is the graph itself, it doesn't seem to represent the numbers in a manner my brain thinks they would. I assumed in the beginning (maybe I still do or don't, not sure anymore) the linear linear line represented and input/output 1:1 ratio. So a 10% force input is 10 out (50 in/50 out etc). Graphically it would seem the TX outputs MUCH higher than the input it recieves starting roughly around a 25% input. But from the input force of 0-10,000 units of whatever, the output of those numbers is MUCH lower. HA!!!!!! Holy sh***.!!!, as I'm typing this out-- Is the line in the graph upside down???? That would also mean the way I explained steep to shallow based on the FFB guide description made more sense to me initially?

poirqc
12-08-2015, 00:35
It's frustrating to see something without being able to explain it clearly.

I'll try

The guide is explained in a certain order.

Damping setup --> Deadzone setup --> Clipping(RA or SC) setup --> Scoops setup

The bloc diagram order is in the opposite way. When we try to find numbers, is the order of settings important or it doesn't matter witch one you setup first?

Let's take a splice of time.

As the force generated move thru the bloc diagram, can it go below 0? In this case, is it kept on so it could go back between 0 and 1 as it advance? When everything is processed, does i still have a value that my wheel can still output?

We ear lots of talk about clipping, but the removal or absence of low signal is as worse. I won't even talk about that, i should start a whole thread on it.

Getting the right numbers to attain the best linearity isn't the end goal, it's the starting point. From there, someone can fine tune to his personnal preferences.

I'm sure there's a lots of info about that on the wmd forum. It's a shame we don't have access to it.

baz00ka
12-08-2015, 00:57
can you guys help me figure out if my understanding of what deadzone removal does is correct or not ? lets say game ffb system outputs to a virtual wheel from say -900 to 0 to 900. now we take a real wheel and due to its mechanical design shortcomings its 'dead' say -5 one way and 5 the other. so when you configure deadzone removal you basically tell the game to 'shift' its ffb output so its starts after 'dead' offsets both ways. what does it really gain you tho ? does it make things less abrupt and more smooth when you transition from deadzone to where the wheel starts to respond ? is that what its for ? why there is another option in game config to then set a deadzone (same screen where where sensitivity is configure), whats that for ? after re-reading an unofficial ffb guide i kinda figured the rest as audio processing units analogy eg soft clipper -> compressor, relative adjust -> sustainer, scoop -> 2 band equalizer so i kinda play with values with that concept in mind and gotten some good results via alot of trial and error but deadzone removal thing i just set it and falloff to zero bc i just couldnt feel any difference with small values and with larger values TX starts acting all weird for me.


Well, conceptualy, since i'm still not sure about numbers, is that deadzone removal should be set precisely. With Deadzone removal, the only goal, imho, should be to only remove the values generated by the game where the wheel can't respond to that.

If you don't remove enough, the game will still generate values the wheel can't output.
If you remove too much, the wheel won't output some fine torques changes you could feel.

P.S.: You owe me a glass of stout! I spilled mine when i tried to spell check! :D

PTG Ty1er Ward
12-08-2015, 01:07
It's frustrating to see something without being able to explain it clearly.

I'll try

The guide is explained in a certain order.

Damping setup --> Deadzone setup --> Clipping(RA or SC) setup --> Scoops setup

The bloc diagram order is in the opposite way. When we try to find numbers, is the order of settings important or it doesn't matter witch one you setup first?

Let's take a splice of time.

As the force generated move thru the bloc diagram, can it go below 0? In this case, is it kept on so it could go back between 0 and 1 as it advance? When everything is processed, does i still have a value that my wheel can still output?

We ear lots of talk about clipping, but the removal or absence of low signal is as worse. I won't even talk about that, i should start a whole thread on it.

Getting the right numbers to attain the best linearity isn't the end goal, it's the starting point. From there, someone can fine tune to his personnal preferences.

I'm sure there's a lots of info about that on the wmd forum. It's a shame we don't have access to it.


Are you sure the force flow chart is in reverse? Anyway, here's how I've been looking at it. For the sake of example and clarity let's say I have an Accuforce wheel setup to perfect 1:1 linearity. If I'm at a dead stand still my FFB input/output are 0. Now lets say I'm in the middle of a high lateral G speed turn, my FFB is giving me the feeling of the grip and let say that's 50% input 50% output. Now let's say I'm running wide and hit some rumbles. Since I still have 50% of the Force Range available the FFB engine can add that feeling to the FFB and I can feel the detail of running over the individual rumbles. If my FFB or Tire Force was set too high and the same turn used 100% FFB, when I hit the rumbles and wouldn't have the same feeling because I'm already at max FFB range and now I'm clipping.

You can never really have a force below zero, the millisecond that your car moves away from a 0 X,Y or Z axis there is information being sent to the wheel, its just a question of how and when your wheel relays the information to your hands. I think the main reason IMHO for striving to achieve the best linearity you can with you given wheel is having a nice FFB strength level while maintaining the ability have fine and subtle details in the FFB.

PTG Ty1er Ward
12-08-2015, 01:28
can you guys help me figure out if my understanding of what deadzone removal does is correct or not ? lets say game ffb system outputs to a virtual wheel from say -900 to 0 to 900. now we take a real wheel and due to its mechanical design shortcomings its 'dead' say -5 one way and 5 the other. so when you configure deadzone removal you basically tell the game to 'shift' its ffb output so its starts after 'dead' offsets both ways. what does it really gain you tho ? does it make things less abrupt and more smooth when you transition from deadzone to where the wheel starts to respond ? is that what its for ? why there is another option in game config to then set a deadzone (same screen where where sensitivity is configure), whats that for ? after re-reading an unofficial ffb guide i kinda figured the rest as audio processing units analogy eg soft clipper -> compressor, relative adjust -> sustainer, scoop -> 2 band equalizer so i kinda play with values with that concept in mind and gotten some good results via alot of trial and error but deadzone removal thing i just set it and falloff to zero bc i just couldnt feel any difference with small values and with larger values TX starts acting all weird for me.

The FFB Deadzone removal basically tells the game at what FFB force level your wheel is actually capable of reproducing a force (from my testing this is between 6-12 with the TX). I suggest you try 0.08 for removal and 0.04 for falloff, or 0.12 and 0.06 depending on your main FFB setting. That's the one conclusive MAJOR improvement these linearity tests have provided.

The Deadzone setting for the wheel (the one paired with sensitivity) you should leave at 0 IMHO if you have a TX wheel. Sensitivity I suggest at 50 (linear) but that's up to you.

It's easier I think to explain Sensitivity with a Pedal first I think, so at 50 means that you have a linear/even input through out the pedal travel (or technically supposed to have) Making the number higher or lower changes the input curve (away from linearity).

Increasing the number (higher than 50) means more input is applied at the beginning of the curve - when you first start to press the pedal

Decreasing the number (lower than 50) means more input is applied at the end of the curve - when you are nearing the end of pedal travel

example- if your are driving a high RWD torque car with no TCS, a LOWER Throttle Sensitivity can "to a degree" help you manage wheel spin because less input is applied at the beginning/early stages of pedal travel.

Steering sensitivity works in the same way but in both directions (away from center) and does get a little more complicated as it is also effected by DOR and Steering Ratio So 0 Deadzone and 50 Sensitivity is a safe bet for wheel users IMHO.

PTG Ty1er Ward
12-08-2015, 01:55
now we take a real wheel and due to its mechanical design shortcomings its 'dead' say -5 one way and 5 the other. so when you configure deadzone removal you basically tell the game to 'shift' its ffb output so its starts after 'dead' offsets both ways. what does it really gain you tho ?

To make it more clear based on your example: sensitivity adjusts the steering input, so your real life steering wheel example of -5/+5 example could be corrected by increasing the Sensitivty (steering input) above 50. How heavy or light or not at all that you FEEL the steering move is FFB deadzone removal. Technically you could have no FFB and the steering wheel still turns the front wheels how quickly or slowly the front wheels react to the steering wheel tuning is sensitivity. Hopefully that's easier than what I initially said :)

poirqc
12-08-2015, 02:19
Are you sure the force flow chart is in reverse? Anyway, here's how I've been looking at it. For the sake of example and clarity let's say I have an Accuforce wheel setup to perfect 1:1 linearity. If I'm at a dead stand still my FFB input/output are 0. Now lets say I'm in the middle of a high lateral G speed turn, my FFB is giving me the feeling of the grip and let say that's 50% input 50% output. Now let's say I'm running wide and hit some rumbles. Since I still have 50% of the Force Range available the FFB engine can add that feeling to the FFB and I can feel the detail of running over the individual rumbles. If my FFB or Tire Force was set too high and the same turn used 100% FFB, when I hit the rumbles and wouldn't have the same feeling because I'm already at max FFB range and now I'm clipping.

You can never really have a force below zero, the millisecond that your car moves away from a 0 X,Y or Z axis there is information being sent to the wheel, its just a question of how and when your wheel relays the information to your hands. I think the main reason IMHO for striving to achieve the best linearity you can with you given wheel is having a nice FFB strength level while maintaining the ability have fine and subtle details in the FFB.

I agree with you that force generated from wheel movement can never be 0. However, as the data moves thru the flow chart, a wrongfully tuned step could(i don't know) put the data back to 0. If i put too much Deadzone removal or to much Scoop reduction.

Logicaly, the flow chart should be read the same way we read english text?

Just had a new look at the flow char. There's arrow between the steps that confirm i was reading it the right way. In my last post, when i talked about "Damping setup --> Deadzone setup --> Clipping(RA or SC) setup --> Scoops setup" I was talking about the order of the text explanation in the beginning of the pdf.

PTG Ty1er Ward
12-08-2015, 02:42
I agree with you that force generated from wheel movement can never be 0. However, as the data moves thru the flow chart, a wrongfully tuned step could(i don't know) put the data back to 0. If i put too much Deadzone removal or to much Scoop reduction.

Logicaly, the flow chart should be read the same way we read english text?

Just had a new look at the flow char. There's arrow between the steps that confirm i was reading it the right way. In my last post, when i talked about "Damping setup --> Deadzone setup --> Clipping(RA or SC) setup --> Scoops setup" I was talking about the order of the text explanation in the beginning of the pdf.

Ahh, yes. We're both reading it the same way. I got confused by you saying "the bloc diagram" being the opposite way. I thought you were reffering to the signal chain flowchart.

"Wrongly tuned FFB" settings creating zero data isn't really a possibility from what I've read, since that any signal compression is fractional (i can't imagine anything in the FFB math dividing by zero) but moving it to an inperceptable level is likely (it's essentially the same thing to driver). I agree with what you're saying though.

I have a feeling the Deazone Removal reconfigures the full FFB range based on your Deadzone setting i.e. setting your dz at 10 squeezes 0-100 force into your 90% range. All the wheelchecker tests at various FFB levels (thereby lowering and raising the dz) seem to support that theory but no way of knowing for sure.

Schnizz58
12-08-2015, 02:59
I think what has turned me upside down is the graph itself, it doesn't seem to represent the numbers in a manner my brain thinks they would. I assumed in the beginning (maybe I still do or don't, not sure anymore) the linear linear line represented and input/output 1:1 ratio. So a 10% force input is 10 out (50 in/50 out etc). Graphically it would seem the TX outputs MUCH higher than the input it recieves starting roughly around a 25% input. But from the input force of 0-10,000 units of whatever, the output of those numbers is MUCH lower. HA!!!!!! Holy sh***.!!!, as I'm typing this out-- Is the line in the graph upside down???? That would also mean the way I explained steep to shallow based on the FFB guide description made more sense to me initially?
No, it's not upside down. What you said in the 1st few sentences is exactly right. Where the curve is steep, you're getting more bang for your buck. For a certain input you're getting more output than 1:1. But then the curve flattens out and it goes the other way. For more input, you get less output. [By the way, by input I mean the force that the computer tells the wheel to apply and by output I mean the force that the wheel actually does apply.] Scoop processing applies a transform that reduces the gain for small forces and increases it for large forces. You put the two together and they cancel each other out to produce a straight-ish line.

Schnizz58
12-08-2015, 03:21
It's frustrating to see something without being able to explain it clearly.

I'll try

The guide is explained in a certain order.

Damping setup --> Deadzone setup --> Clipping(RA or SC) setup --> Scoops setup

The bloc diagram order is in the opposite way. When we try to find numbers, is the order of settings important or it doesn't matter witch one you setup first?
It doesn't matter what order you setup the configuration stuff. The only things that matter are the final values. The feedback signal flows through the block diagram, left to right, and each block of processing uses whatever parameters are appropriate for that block. The scoop function uses scoop reduction and scoop knee for example. None of the blocks really know anything about the other blocks. They each just take the signal in their inputs, do whatever processing they're programmed to do and then hand the output off to the next block.



Let's take a splice of time.

As the force generated move thru the bloc diagram, can it go below 0? In this case, is it kept on so it could go back between 0 and 1 as it advance? When everything is processed, does i still have a value that my wheel can still output?
It doesn't really matter how it's represented internally because it's going to get scaled at the end by the FFB scale into a range of values that the wheel can understand. One way of doing it would be that positive values represent a clockwise torque and negative values represent a counter-clockwise torque.


We ear lots of talk about clipping, but the removal or absence of low signal is as worse. I won't even talk about that, i should start a whole thread on it.
Yeah, I've found that to be a tricky one. At first I had too much deadzone removal and then I turned it off altogether. Now I have a little bit with a gradual falloff.



Getting the right numbers to attain the best linearity isn't the end goal, it's the starting point. From there, someone can fine tune to his personnal preferences.
Absolutely. It's just like tuning a car. You can use Jussi's spreadsheet to get pretty close but you still have to drive it.

Schnizz58
12-08-2015, 03:35
I agree with you that force generated from wheel movement can never be 0.
Force can be 0. If you turn the wheel from one side to the other you have to pass through a point where the force is 0.


However, as the data moves thru the flow chart, a wrongfully tuned step could(i don't know) put the data back to 0. If i put too much Deadzone removal or to much Scoop reduction.
I don't think so. Deadzone removal only adds force, not reduce it. Scoop reduction reduces it but it can never reduce it to zero.


Logicaly, the flow chart should be read the same way we read english text?
Yes, the inputs are on the left and the output is on the right (except for the g force thing in the middle).

Schnizz58
12-08-2015, 03:39
I have a feeling the Deazone Removal reconfigures the full FFB range based on your Deadzone setting i.e. setting your dz at 10 squeezes 0-100 force into your 90% range.
That would be an interesting way to do it but that isn't my understanding (however I'm willing to be wrong). I think it boosts the signal when it's small and does nothing beyond the falloff range. How much it boosts it, I don't know. If it worked as you described, it wouldn't need a falloff parameter.

PTG Ty1er Ward
12-08-2015, 04:15
Okay, here's maybe something else to consider. Here are the results of a step log 2 with the 4 averages being comprised of FFB at 40-60-80-100 and the existing Spreadsheet formula are being used.

214901



Here are the exact same input/output logs but with the Math done differently (like how Schnizz pointed out 10,000 input units of whatever but his wheel outputs 7500. Mine actually does ~4800 at 10K and so does Inthebagbud's, my wheels' got 1000 hours on it so that's the culprit I guess). Based on those numbers, doesn't this one make more sense in that regard? Especially in in terms of Scoop. Forces below the knee are reduced (x-%) and above the knee they become input as 1 to output as 1 (I still think they are scaled to 1 if they are below 1)

214902



I would put the Knee around 65 with a 20

PTG Ty1er Ward
12-08-2015, 04:43
That would be an interesting way to do it but that isn't my understanding (however I'm willing to be wrong). I think it boosts the signal when it's small and does nothing beyond the falloff range. How much it boosts it, I don't know. If it worked as you described, it wouldn't need a falloff parameter.

Good point. It's possible that "falloff" may actually be the "boost" factor, which is why 4:1 and 2:1 seem to be the best setting. 4:1 at a dzr of 16 and 2:1 at a dzr of 8. Falloff meaning falloff towards zero instead of falloff into the increase in input. That would be an odd thing to call it (falloff) if that's the case, but hey, just throwing ideas out there

poirqc
12-08-2015, 05:55
Good point. It's possible that "falloff" may actually be the "boost" factor, which is why 4:1 and 2:1 seem to be the best setting. 4:1 at a dzr of 16 and 2:1 at a dzr of 8. Falloff meaning falloff towards zero instead of falloff into the increase in input. That would be an odd thing to call it (falloff) if that's the case, but hey, just throwing ideas out there

It may be like that. I did a test a while back (http://forum.projectcarsgame.com/showthread.php?35635-G29-Discussions-and-Support&p=1064827&viewfull=1#post1064827). I didn't try to read precise data value on the screen tho.

It would be something to do.

In essense, it seems the falloff parameter is also doing some build up forces. I still think you need to first put the DRR first, then the DRF according to the DRR. The lower the falloff is, the stiffer the wheel get, and you get less small details. The higher DRF gets, the lighter the wheel is, but you can feel smaller details more.


I have a feeling the Deazone Removal reconfigures the full FFB range based on your Deadzone setting i.e. setting your dz at 10 squeezes 0-100 force into your 90% range. All the wheelchecker tests at various FFB levels (thereby lowering and raising the dz) seem to support that theory but no way of knowing for sure.


The deadzone removal range is the big FFB deadzone block and the falloff, with a 1/10th of a value, is the last precision part of it.

With that said, even if i think i understand Deadzone removal by it's own, i'm not sure how it's applied to the graph.

214904

Does it move the entire curve or it add force only to the first part that gets no output. If so, it's the later, then we have some kind of crossover distortion where a small zone will be again output less.

inthebagbud
12-08-2015, 08:04
Damm I go out for a night and see what happens, that's some catching up I have to do

I have done a skim read and it seems to me that unless we know which Pcars settings are the equivalent to the TX CP settings we are struggling to provide a definitive set of data

I think we all agree that FFB = Overall Strength

Constant could be = to TF from post http://forum.projectcarsgame.com/showthread.php?35147-Linear-FFB-Test-Excel-sheet-Universal&p=1080592&viewfull=1#post1080592


214907

Lars Rosenquist
12-08-2015, 12:23
Does it move the entire curve or it add force only to the first part that gets no output. If so, it's the later, then we have some kind of crossover distortion where a small zone will be again output less.
I always figured it works like this (G27 as an example):

if (input < 0.16) then output = 0.16 else output = input

and not like this:

output = input + 0.16

Can we test this in WheelCheck? Should be observable if it is.

PTG Ty1er Ward
12-08-2015, 13:56
I always figured it works like this (G27 as an example):

if (input < 0.16) then output = 0.16 else output = input

and not like this:

output = input + 0.16

Can we test this in WheelCheck? Should be observable if it is.

Yes, this can be tested in WheelCheck (I also see where the 4:1/2:1 falloff comes from). Major update coming at some point today. I just remebered the iRacing forum isn't public (I kind of assumed up to this point everyone else was going there already for the info). It's a lot to relay/explain but I'll try to condense it in a manner hopefully easy to understand for all. Some things we've been doing right, some things we haven't (depending on your wheel).

poirqc
12-08-2015, 15:23
I always figured it works like this (G27 as an example):

if (input < 0.16) then output = 0.16 else output = input

and not like this:

output = input + 0.16

Can we test this in WheelCheck? Should be observable if it is.

If it would be like: if (input < 0.16) then output = 0.16 else output = input, the only value possible below 0.16 would be 0.16. It would only offset the deadzone. The transition from 0 to 0.16 would be too sharp. Falloff wouldn't help because the wheel couldn't output the smoothing.

I think.


Yes, this can be tested in WheelCheck (I also see where the 4:1/2:1 falloff comes from). Major update coming at some point today. I just remebered the iRacing forum isn't public (I kind of assumed up to this point everyone else was going there already for the info). It's a lot to relay/explain but I'll try to condense it in a manner hopefully easy to understand for all. Some things we've been doing right, some things we haven't (depending on your wheel).

Indeed, it's closed. I was too lazy to try a test subsciption.

Schnizz58
12-08-2015, 15:40
I always figured it works like this (G27 as an example):

if (input < 0.16) then output = 0.16 else output = input

and not like this:

output = input + 0.16

Can we test this in WheelCheck? Should be observable if it is.
I think it's more like

if (input < DRrange)
output = boost * input;

If the input is in the falloff range less gain is applied, depending on where in the falloff zone the input falls.

PTG Ty1er Ward
12-08-2015, 17:17
I'm not sure how or where to start, so since DZR and DZF are the current topic here goes. Since we're using WheelChecker I'll reference info from that the guy the actually wrote the utility (who is also the iRacing FFB designer/engineer) David Tucker and Clayton Macleod who seems to be the resident FFB expert/Guru. Anything in "quotes" is direct, not in quotes is my 2 cents.

iRacing FFB (and wheelchecker) refer to a Minimum Force Value setting which I believe equals PCars Deadzone Removal Range setting. Deadzone Falloff is adjustable is PCars, in iRacing it's hardcoded at 25% (4:1). Here's the iRacing explanation:

214943

""If I follow that line down to the left, it seems to be pretty linear until it gets down to about the 1300 mark. Below that the response is noticeably less than perfect, right down to the 1000 mark where it first shows movement. So let's say I take that 1300 value as my minimum force value. 1300 / 4 is 325, so FFB values from 0 to 325 coming from the car will actually get scaled to between 0 and 1300 when sent to the wheel. And then values of 326 and over coming from the car are scaled from 1301 to 10,000 when sent to the wheel. This should provide a smoother transition from 0 up through the minimum force my wheel is actually capable of, 1000, and then beyond that to where the wheel becomes more or less linear beyond 1300. Your wheel will likely vary, so run the step2 test yourself, as outlined further down. I would say this is now the preferred method, and we should ignore the min. force test in wheelcheck for now.

We are likely to see minimum force values roughly 25% larger with this method, In my case, 1300 is in the general neighbourhood of 25% higher than I was previously using.""

Here's my TX output graph using the exact WheelChecker/iRacing TMPC suggestions:

214944

konnos
12-08-2015, 17:54
Actually the Minimum Force Test cannot be trusted to determine DRR in project Cars. It's better to use the Linear Force Test values (as seen in the above graph) and see when forces begin to become felt and use that as your % for deadzone removal.

Min Force test says 0% for my DFGT, but Linear force says 3-6% (there's a small flat line in between those numbers and you can choose to ignore it). And it does feel better when i up the DRR so test it and see.

PTG Ty1er Ward
12-08-2015, 18:04
Actually the Minimum Force Test cannot be trusted to determine DRR in project Cars. It's better to use the Linear Force Test values (as seen in the above graph) and see when forces begin to become felt and use that as your % for deadzone removal.

Min Force test says 0% for my DFGT, but Linear force says 3-6% (there's a small flat line in between those numbers and you can choose to ignore it). And it does feel better when i up the DRR so test it and see.

Agreed- the minimum force test on my TX was 1%. It does go on to mention that in the current version of wheelchecker the step2 data is better to use (used in my graph) instead of min force test. The min force checker applied in older versions of iRacing when there was no "Falloff" equivalent and it was 0:0 10,000-10,000

P1ckN1cker2406
12-08-2015, 18:22
So du you have new Settings for the TX Wheel after your latest Test?

PTG Ty1er Ward
12-08-2015, 18:30
So du you have new Settings for the TX Wheel after your latest Test?

Yes, for PC. Not for Xbox (yet). Now it's on to getting the FFB the same way without the benefit of TMCP, but based on things I've just read confirming some of my original thoughts I'm hoping it will be fairly straightforward.

P1ckN1cker2406
12-08-2015, 19:00
Ok cool. Can you tell me your PC Settings, because I play Project Cars on both Plattforms too. :)

And what new Infos do you just read about the Xbox?

Sry for asking so much. But Im very interested on the FFB System of Project Cars and I did a Wheel Setup Guide on a very popular German Sim Racing Forum for Xbox One. ;)

PTG Ty1er Ward
12-08-2015, 20:20
My TMCP settings are 45- 100/100/100/100 (see the graph above). The DZ is still very reasonable there, so I'm set at DZR 0.12/0.03 DZF (4:1)

I'm not finished with the full in-game setup on PC, but right now it's FFB at 100% No Scoop, Relative Adj or Soft Clipper. From here (for me personally) I think it's more about Tire Force and the Spindle Arm to tidy things up but we'll see.

It's not something that I read about Xbox, but something I read that confirmed what I had been thinking for a while. I won't start testing things out on XBox until it gets 3.0 (same new FFB options as PC 2.5 currently). Here's the iRacing explanation of why to leave the Damper at 100% in TMCP, but the same rule applies to all 5 settings (and it remains to be seen if PCars is the same I can't conceivably understand why it would not be):

"The reason for setting 100% dampening in your wheel's drivers is that the sim's setting is a percentage of what you have set in the drivers. So if you have 0% set in the sim, it doesn't matter what setting you have in your drivers. But if you set 0% in your drivers, and then decide you want to try the old sim default of 75%, for example, then you won't feel any difference. After all, 75% of nothing is nothing. So in order for the damping slider in the sim to have any affect, you must have it turned on in the drivers. If you have the drivers set for less than 100%, let's say 50%, then the sim's setting will be a combination of the two. 50% in the drivers and 50% in the sim means an actual total of 25% damping."

So when I say the first thing I going to try on Xbox 3.0 is the Main in-game FFB setting at 60% it because without knowing for sure my guess is the the Xbox driver is the same default as PC (at 75- 100/100/100/100) and 75 is to 100 as 45 is to 60.

It's funny because I've seen bunches of posts in dozens of forums for half a dozen PC sims/racers with people all saying that TM's Main FFB slider shouldn't be set higher than 60 because it starts internally "boosting" the signal at that point. I didn't pay it much attention because at the time the only Xbox use it got was Forza, and for iRacing and Assetto I just used whatever TMCP everyone generally seemed to say was the best way (just like I originally used Bmanic's for PCars). But one problem is the PC community doesn't know/understand that the TMCP settings don't have any effect on XBOX (being PC guys though if they actually thought through the way a hardware driver works they would get it). Maybe Fanatec and/or Logitech actually save changes to an internal memory (I don't know) so they assume TM does it to. So XBox people see FFB posts for TM gear that start with TMCP settings at the core and then "must follow everything exactly" not understanding that TMCP has no relevance to XBOX and you don't have the same slider granularity on the box. Sometimes it feels like half the XBOX threads here are wheel/FFB related issues and as console users they generally just want to pick up and play, not mess around with FFB settings (bringing back "Classic" in 3.0 will satisfy a lot of people I think)

Since PCars released I really haven't wanted to play anything else and it's the first cross platform Sim, sowith all the changes to the FFB system, half the time it seems like the PC/XB have been on different patch levels (NOT a complaint keep them coming SMS :)). I noticed I was spending a lot time looking for and trying "new" FFB settings for one platform or the other, so I just figured it would be better for me in the long run to just learn the system. I intend on running PCars for a very long time and I'm also a Gold Member PCars2. So, here I am and that's where I'm at.

P1ckN1cker2406
12-08-2015, 20:27
Thx for the very good explanation.

Yeah Im member on PCars2 too. :D

poirqc
12-08-2015, 22:56
First, i want to thank you PTG Ty1er Ward for taking the time to check how wheelcheck is supposed to work.

From the first part of it, it seems we where right about using the linear force test to find the deadzone of a wheel instead of the min force test.

However, this giving weight to my theory that the driver setup is the starting bloc how a wheel behave. After that, the behavior of the wheel will change according to the software that runs on top of it.

Wheelcheck will output a signal that is unadulterated. It's easy to interpret. iRacing's developpers explained how they compute the value. It's again easy to interpret. Right now Deadzone Removal is easy to interpret, but we only have thoeries for it. There's even more theories about Scoops. Lots of them make sense.

However, will still need to figure the relation between Deadzone Removal and Scoops.

Sadly we don't have anything to test if our calculations are good. Test drives is for sure a good one, but it can get confusing and hard to track down at some point. Even knowledgable members like Lars still have some questions.


I'm starting to think we should open a repository thread where everyone can put his controller config. Even if this would be blind data gathering, everyone could benefit from everyone's testing.

Right now, it's still hazy.

Cheers!

baz00ka
12-08-2015, 23:34
thanx man! that makes sense now however something here isnt quite right. so 'deadzone removal' (kinda misleading name for this if you ask me because of the other deadzone but i digress) is basically a signal booster/scaler but instead of scaling the whole thing it lets you define where that boost starts and where it ends with 'deadzone removal falloff'. i couldnt try 0.08/0.04 values bc i'm on xbox and sliders are screwed up there living their own life but did try 0.12/0.06 and its too much for me. what i ended up doing is 0.10/0.10 and in fact if i could bump falloff any further i would. the reason being falloff is either not coded properly or bad idea to begin with. the reason being deadzone removal is responsible for that awful wheel side to side rocking effect. i think what should happen is a single parameter to re-scale the whole ffb signal from a configurable value eg if the wheel starts to respond from value X then rescale the whole thing to match that. instead what seems to be happening is it boosts the signal to clear the deadzone but the rest of it is not scaled to match that initial boost range and it must be creating an abrupt transition between boost range and everything else. this must be what falloff is supposed to smooth out and maybe it does but it does not do it well enough to avoid that awful rocking effect/artefact, it can only minimize it to certain degree. my take away from this is right now with patch 2.0 it is best to set the falloff to its maximum and definitely not less than range. it could be some of my other ffb settings play into this but if i understand this correctly now i think its plain broken.


The FFB Deadzone removal basically tells the game at what FFB force level your wheel is actually capable of reproducing a force (from my testing this is between 6-12 with the TX). I suggest you try 0.08 for removal and 0.04 for falloff, or 0.12 and 0.06 depending on your main FFB setting. That's the one conclusive MAJOR improvement these linearity tests have provided.

skoader
12-08-2015, 23:36
...I'll post details later today, we really need to be on the same page here before moving forward.

I think some confusion has arisen because of the way poirqc and I scale the data which allows us to place force curves and wheel response curves into the same graph.

Without the appropriate scaling, the force curves (linear or scooped) should in a seperate graph to the wheel response curves because the y values have a completely different meaning.
I think it might be useful to seperate and take a look at them individually. Please bear with me, as I'm attempting to be as clear as possible.

First, a scooped force graph -

x axis = input force
y axis = output force

214964

This graph represents a scoop function set up with the given parameters. You can look up an input force on the x axis and retrieve an output force on the y axis.

For example -
An input force of 0.54 (x axis) is 'scooped' and is output as 0.30 (y axis).
An input of 0.80 is output as 0.70 etc. The dashed line represents linear force.


But for the wheel response graphs -

x axis = input force
y axis = wheel response - This can be expressed in degrees[0-450](depending on the wheel), position values [0-10000] or can even be normalized.

214965 214966

These graphs show the wheels response as a function of force - You can look up an input force on the x axis and retrieve a wheel response on the y axis.

For example -
Given a force of 0.5 (x axis) the wheel responds by rotating 293 degress (y axis on first graph) or to a position of 6516 (y axis on the 2nd graph)

Without the appropriate scaling, inserting scooped or linear force curves into these graphs isn't correct. The y axis in both cases represents wheel movement and force curves cannot be accurately expressed.
This is pretty clear in the first graph where the y axis represents degress of rotation. Because wheelcheck is outputting raw Direct Input force values [0-10000] on the x axis it might be tempting to do it on the second graph because the axes are scaled the same but the problem still stands...

For example -
A straight line moving from the bottom left to the top right on either wheel response graph is not an accurate respresentation of linear force. Such a line only shows a linear response from the centre all the way to full lock. Forces are only expressed on the x axis in both of these graphs.

To put them together correctly requires that we normalize the data so that we have a common scale on the y axis.

Note that normalizing the response values is fine for analyzing a single response curve. It isn't a good idea if you want to compare response curves with different driver settings. The differences in maximum reported rotation is useful information in this context and it will be lost if the data is normalized.


So how do we calculate a wheels response to scoop? The graphs I posted here show what is required.

1. Run our input forces[0-1] through the scoop function. Graph 1.
2. Plug the resulting scooped forces into one of the wheel response functions. Graph 2 or 3.

We can't just scale the wheel response values, because the those responses are nonlinear and you won't get the correct results.

Unfortunately, I don't have a simple equation that will help. I jumped into the toolbox I know and do it programmatically using a lookup table of response values with some helper functions to correctly interpolate between those values as needed. I can then do a simple lookup to get an accurate wheel response for any force I need.

I hope that was reasonably clear and I haven't muddied the waters even further.

inthebagbud
13-08-2015, 00:59
thanx man! that makes sense now however something here isnt quite right. so 'deadzone removal' (kinda misleading name for this if you ask me because of the other deadzone but i digress) is basically a signal booster/scaler but instead of scaling the whole thing it lets you define where that boost starts and where it ends with 'deadzone removal falloff'. i couldnt try 0.08/0.04 values bc i'm on xbox and sliders are screwed up there living their own life but did try 0.12/0.06 and its too much for me. what i ended up doing is 0.10/0.10 and in fact if i could bump falloff any further i would. the reason being falloff is either not coded properly or bad idea to begin with. the reason being deadzone removal is responsible for that awful wheel side to side rocking effect. i think what should happen is a single parameter to re-scale the whole ffb signal from a configurable value eg if the wheel starts to respond from value X then rescale the whole thing to match that. instead what seems to be happening is it boosts the signal to clear the deadzone but the rest of it is not scaled to match that initial boost range and it must be creating an abrupt transition between boost range and everything else. this must be what falloff is supposed to smooth out and maybe it does but it does not do it well enough to avoid that awful rocking effect/artefact, it can only minimize it to certain degree. my take away from this is right now with patch 2.0 it is best to set the falloff to its maximum and definitely not less than range. it could be some of my other ffb settings play into this but if i understand this correctly now i think its plain broken.

Completely agree - I cannot use DRR or DRF at all as the rocking is too much to cope with

As I have said before we do not know what the in game defaults actually represent and if for arguments sake on the Xbox they are actually based on the TX wheel and SMS have optimized for the TX then using the FFB settings to change may be counter productive.

We can do as much testing and theorising but without a baseline to work from we cannot reach any definitive conclusions on the xbox

The PC yes conclusions can be reached as you can alter the wheel control panel, test this and use directly in the game and know what you should expect all we can do in xbox is try and understand what the sliders do, use them and if we don't get the result we thought we should - move on to something else and test that.

I have been through so many different combinations of settings and can honestly say I can get same driving feel just by using different degree of FFB and TF as I can using TF and any combination of Knee, RAG or Clip

That is until the dreaded oscillation appears like just now aargh - which tells me it is time to stop testing

For Xbox users the idea of poirqc of a repository thread would be good idea, but with the changes in patch 3 we may have to start the debate all over again

If anybody ones to test my settings are

FFB100
TF52
WPS default 0.04
RAG1.38
RAB0.30
RAC1.32

All other DRR, Scoop, Clip are set at zero

Deadzones zero
Sensitivity 50

It is a bit springy and needs getting used too - hopefully the patch 3 will help as they are introducing something for this I seem to remember

Tested mainly with the RGT 8 GTS on Oulton International - time around 1.35 - think I have had it just under with stock car settings and stock car FFB

I think I am going to leave the game FFB alone until patch 3 and concentrate on the car FFB as this is what is suggested as a starting point in the FFB document

The input from everybody has been good and I take my hat off to PTG, Poirqc, skoader and schnizz for their knowledge and time spent helping others

PTG Ty1er Ward
13-08-2015, 01:36
I think some confusion has arisen because of the way poirqc and I scale the data which allows us to place force curves and wheel response curves into the same graph.

But for the wheel response graphs -

x axis = input force
y axis = wheel response - This can be expressed in degrees[0-450](depending on the wheel), position values [0-10000] or can even be normalized.

For example -
A straight line moving from the bottom left to the top right on either wheel response graph is not an accurate respresentation of linear force. Such a line only shows a linear response from the centre all the way to full lock.

I"m going to go ahead and bow out at this point fellas- and I sincerely appreciate everyone's help/information and feedback Big UPs to Poirqc for stating this thread. From what I've read recently Scoops were implemented primarily to account for Logitech Wheels so I don't want to to hinder your pursuit of happiness.

I do want to point out thought that test we've all been using is called a Linear Step Test (the Linear Force test is different and output a different set of data). I believe Skoader is correct that the data for the step 2 test is not conducive on it's own for determining Scoop.

When you say "For example -A straight line moving from the bottom left to the top right on either wheel response graph is not an accurate respresentation of linear force. Such a line only shows a linear response from the centre all the way to full lock" Technically that's correct, but that not what the test is designed to do anyway. It designed to show you how your wheel acts to "Linear Steps" and it's specifically designed to test the linearity of DRIVER settings, "not wheel force linearity response" in the manner your describing. So you might be right this may or may not be right test for determining Scoop application.

If driver Wheel Linearity is your goal than testing the Linear Step response of the Driver is still a very accurate picture for adjusting the initial linear capabilities of the wheel FFB as a starting point. Your second pictures are titled "wheel response function" should really be called "driver response function" because changes you make to the driver drastically effect that curve in many ways. The DOR is part of the step test linearity response. I added a linear step line. With an input force of 5000, if your driver was linear your wheel would output 225 degrees at that step, but it's roughly 300 degrees at that linear step. At input force of 9000 you should have 405 degrees at that linear step but it almost 90 degrees (a full quarter turn) less than that. Our shared common goal I believe is to reduce/flatten the curve while sacrificing the least amount of strength and detail. I just think we can both achieve that in different ways, and hopefully along the way the things people learn all benefit the community as a whole.


214971

PTG Ty1er Ward
13-08-2015, 02:10
thanx man! that makes sense now however something here isnt quite right. so 'deadzone removal' (kinda misleading name for this if you ask me because of the other deadzone but i digress) is basically a signal booster/scaler but instead of scaling the whole thing it lets you define where that boost starts and where it ends with 'deadzone removal falloff'. i couldnt try 0.08/0.04 values bc i'm on xbox and sliders are screwed up there living their own life but did try 0.12/0.06 and its too much for me. what i ended up doing is 0.10/0.10 and in fact if i could bump falloff any further i would. the reason being falloff is either not coded properly or bad idea to begin with. the reason being deadzone removal is responsible for that awful wheel side to side rocking effect.


Completely agree - I cannot use DRR or DRF at all as the rocking is too much to cope with

That is until the dreaded oscillation appears like just now aargh - which tells me it is time to stop testing

The input from everybody has been good and I take my hat off to PTG, Poirqc, skoader and schnizz for their knowledge and time spent helping others

Thanks inthebagbud :) The stuff above though is the main reason why I'm going off the FFB radar for while- I don't want a half-baked idea or an FFB WIPs causing people any problems. I'll resurface soon with with some idea's the for the XBOX, but it will be after 3.0 releases. I'll just keep fiddling on PC until then

poirqc
13-08-2015, 02:47
I"m going to go ahead and bow out at this point fellas- and I sincerely appreciate everyone's help/information and feedback Big UPs to Poirqc for stating this thread. From what I've read recently Scoops were implemented primarily to account for Logitech Wheels so I don't want to to hinder your pursuit of happiness.

I do want to point out thought that test we've all been using is called a Linear Step Test (the Linear Force test is different and output a different set of data). I believe Skoader is correct that the data for the step 2 test is not conducive on it's own for determining Scoop.

When you say "For example -A straight line moving from the bottom left to the top right on either wheel response graph is not an accurate respresentation of linear force. Such a line only shows a linear response from the centre all the way to full lock" Technically that's correct, but that not what the test is designed to do anyway. It designed to show you how your wheel acts to "Linear Steps" and it's specifically designed to test the linearity of DRIVER settings, "not wheel force linearity response" in the manner your describing. So you might be right this may or may not be right test for determining Scoop application.

If driver Wheel Linearity is your goal than testing the Linear Step response of the Driver is still a very accurate picture for adjusting the initial linear capabilities of the wheel FFB as a starting point. Your second pictures are titled "wheel response function" should really be called "driver response function" because changes you make to the driver drastically effect that curve in many ways. The DOR is part of the step test linearity response. I added a linear step line. With an input force of 5000, if your driver was linear your wheel would output 225 degrees at that step, but it's roughly 300 degrees at that linear step. At input force of 9000 you should have 405 degrees at that linear step but it almost 90 degrees (a full quarter turn) less than that. Our shared common goal I believe is to reduce/flatten the curve while sacrificing the least amount of strength and detail. I just think we can both achieve that in different ways, and hopefully along the way the things people learn all benefit the community as a whole.


214971

I think the linear force test is a good indication of the force a wheel can output. Do the test while holding the wheel. As the test goes on, you'll fell the wheel is appling torque more and more. Even if the wheel doesn't move, you'll fell it more. (Don't try that too often, it doesn't fell or sound good). In the end, you'll get the max lateral torque the wheel can generate.

At that point, the only thing we can do to change this behavior is shaping the forces below that maximum output. Since the maximum output is already reached, we can only lowers the values in a certain ways.

Here's my latest sheet. (https://docs.google.com/spreadsheets/d/1gBiV_mNizlVGKXdWCe-VTduiBQIr0LEufTkRbOy7MIk/edit#gid=0) There's lot of cleaning to do there. I wish i was better at maths.

I would like to update de first post, but as PTG Ty1er Ward said, there's still some thinking to do.

What do you think?

baz00ka
13-08-2015, 11:37
thanx guys! waiting for patch 3.0 to land before doing more tweaking and guessing sounds like a good idea. if it doesnt address ffb issues at least maybe it will fix sliders on xbox (iirc from reading release notes) bc they are a bloody mess right now. the only thing i've seen on par with that was Windows Vista progress bars going backwards ;)


Thanks inthebagbud :) The stuff above though is the main reason why I'm going off the FFB radar for while- I don't want a half-baked idea or an FFB WIPs causing people any problems. I'll resurface soon with with some idea's the for the XBOX, but it will be after 3.0 releases. I'll just keep fiddling on PC until then

skoader
13-08-2015, 23:22
I do want to point out thought that test we've all been using is called a Linear Step Test (the Linear Force test is different and output a different set of data)
I'm using v1.72. The log2 test is called the linear force test. I'm pretty sure that's the one everyone has been using.


The DOR is part of the step test linearity response. I added a linear step line. With an input force of 5000, if your driver was linear your wheel would output 225 degrees at that step, but it's roughly 300 degrees at that linear step. At input force of 9000 you should have 405 degrees at that linear step but it almost 90 degrees (a full quarter turn) less than that.
In the log2 test, forces are applied as an impulse. 10000 units of force applied over 300ms does not mean the wheel should rotate to full lock.

SeeNoWeevil
14-08-2015, 11:39
FFB calibration in pCars is so incredibly frustrating for me. So many options, with so little understanding or explanation. I need to know exactly how something works and *why* someone would change it so I can configure it myself. Even the Thrustmaster control panel makes little sense. I think I'm just going to have to find settings that feel somewhat good and then just leave it alone. I've spent so long tweaking things in pCars (controllers, visuals, FFB) it's a little ridiculous.

PTG Ty1er Ward
14-08-2015, 20:29
I'm using v1.72. The log2 test is called the linear force test. I'm pretty sure that's the one everyone has been using.


In the log2 test, forces are applied as an impulse. 10000 units of force applied over 300ms does not mean the wheel should rotate to full lock.

Agreed. Here's a little more info on it:

"Please keep in mind that this data is only showing you how well a wheel performed on a specific test. It can not measure the quality of the wheel, or even how well the wheel will feel when driving. But it is a starting point for comparing the relative potential of various wheels.

For now I have two tests:

1) Test One centers the wheel and then commands it to turn left/right/left at full force, with 300ms delay between steps. Assuming that the wheels all weigh about the same and have a similar amount of drag, this gives us a measure of how much force the wheel can output. You can also measure the latency by taking the difference between the commanded motion and the first recorded motion. I sample the data at 1ms resolution so I can also measure the update rate of the driver at the same time.

Notes:
- Not all wheels weigh the same, so this is not very scientific (TYLER- I did notice the difference between the GTE Rim (heavier) and the F1 Rim (lighter)
- Not all wheels were tested on a machine that can update with a 1ms accuracy, there results may be off by a few ms. (TYLER- in another section it's suggests setting the "Sleep Time" to 10ms instead of 1ms for Thrustmaster)

2) Test Two tries to measure the linearity of a driver. I run the same step test as above, but I do it many times using different power outputs and measure the amount of rotation achieved at the end of 300 ms. By running this test multiple times using different gain settings in the driver I can measure the effect that the sliders in the driver have on the force output of the wheel."

The Step Log 2 test 100% nails the DZR and DZF, it's absolutely perfect for me. In a sheer Cosmic Coincidence, I actually originally used the Step Log 2 test for it's intended purpose(unknowingly at the time), which is seeing the difference in the data after running multiple tests at different driver values. With that being said, if you are also using the data to try an identify other ways of adjusting the FFB- if your in-game Damper setting is Zero you should set your Damper value to Zero in your driver (TMCP or Profiler) before running the tests.

PTG Ty1er Ward
14-08-2015, 21:13
So many options, with so little understanding or explanation. I need to know exactly how something works and *why* someone would change it so I can configure it myself. Even the Thrustmaster control panel makes little sense.


The Constant and Periodic are static effects and they are not linked directly to the wheel position or wheel velocity. The Spring and Damper are dynamic effects, because they rely on information which depends on the wheel- position for spring or speed for damper.

Here's what each slider in the TMCP does/represents:

Constant: A constant force will keep the same level in time. When the game decides to apply a force of x% of what the wheel can do, the "Constant" will keep that force at the same level. Example: A constant force can be used to simulate your tires' grip level. Depending on the speed and the wheel angle, the force can/will be increased accordingly, or reduced in the event of loss of grip.

Periodic: A periodic force will vary in time according to the type of effect, strength and frequency. Because a periodic force includes an offset that has the same behavior as a constant force, the FFB engine can create a periodic force and merge a feeling produced as a constant force (like tire grip) and a periodic effect (like driving over Rumbles/Curbs). Example: A periodic effect gives the shaking effect on the wheel. At high frequency, you will feel it rumble, but in some cases very a low frequency can be used to a create crash effects, like hitting another car or a wall

Spring: A spring force is a force that increases according to how far you are from a specific position on the wheel. The basic spring force can be a default spring center where the wheel goes back to its center position when you release it, but the force can be set to negative, which will make the wheel go further away from the center position.

Damper: A damper force controls how the wheel will react when it's moving. Moving in one direction it can be used as a dynamic friction (like wheel drag) or in the opposite direction while the wheel gets back to its center position (like a spring effect). It's usually used in order to make it harder to turn the wheel while in other circumstance make it feel like if it's very smooth and easy. Example: I think the PCars description is that damper effects the "Weight of the Wheel", so same thing. Harder to turn would mean a heavier feeling


Edit- this is really just a generic explanation, but it's doubtful the PCars FFB engine uses them in a different manner, albeit in a more advanced way

SeeNoWeevil
15-08-2015, 10:32
The Constant and Periodic are static effects and they are not linked directly to the wheel position or wheel velocity. The Spring and Damper are dynamic effects, because they rely on information which depends on the wheel- position for spring or speed for damper.

Here's what each slider in the TMCP does/represents:

Constant: A constant force will keep the same level in time. When the game decides to apply a force of x% of what the wheel can do, the "Constant" will keep that force at the same level. Example: A constant force can be used to simulate your tires' grip level. Depending on the speed and the wheel angle, the force can/will be increased accordingly, or reduced in the event of loss of grip.

Periodic: A periodic force will vary in time according to the type of effect, strength and frequency. Because a periodic force includes an offset that has the same behavior as a constant force, the FFB engine can create a periodic force and merge a feeling produced as a constant force (like tire grip) and a periodic effect (like driving over Rumbles/Curbs). Example: A periodic effect gives the shaking effect on the wheel. At high frequency, you will feel it rumble, but in some cases very a low frequency can be used to a create crash effects, like hitting another car or a wall

Spring: A spring force is a force that increases according to how far you are from a specific position on the wheel. The basic spring force can be a default spring center where the wheel goes back to its center position when you release it, but the force can be set to negative, which will make the wheel go further away from the center position.

Damper: A damper force controls how the wheel will react when it's moving. Moving in one direction it can be used as a dynamic friction (like wheel drag) or in the opposite direction while the wheel gets back to its center position (like a spring effect). It's usually used in order to make it harder to turn the wheel while in other circumstance make it feel like if it's very smooth and easy. Example: I think the PCars description is that damper effects the "Weight of the Wheel", so same thing. Harder to turn would mean a heavier feeling


Edit- this is really just a generic explanation, but it's doubtful the PCars FFB engine uses them in a different manner, albeit in a more advanced way
Thanks for that. Why does everyone recommend turning some of these to zero? Most say spring and damper to zero, some also say periodic. Why would I want the driver to ignore certain signals from the game?

STEELJOCKEY
15-08-2015, 12:55
Certainly been an interesting read, and doing the test backs up what I've been feeling. Very handy though to pinpoint DZR and where scoop knee should be and how much reduction to apply.

215142

This is what came out for my test.

I used it and applied DZR of 0.06, using the 4:1, I put in DZF of 0.015, put the knee in at about 0.42 (0.46 actually due slider difficulties), and reduction of 0.44 or thereabouts (due sliders again).

I must say I'm liking the outcome, with all other sliders at default.

poirqc
15-08-2015, 13:24
Hi guys, i've cleaned up the sheet and i'm about to post and updated version.

However, i'm stuck. I would like to be able to put and easy Scoop calculation fonction. Right now it's done manually. I need to generate the data in 2 steps. The shallow part and the steep part. I want to be able to generate the data easily for anyone who put's his own data.

Right now, the only way i was able to do it was caculate the steps then write them manually drag the data in sheet.

I was able to do mine data manually, but it may lot's of work for someone who never played with the sheet.

Here's the sheet. (https://docs.google.com/spreadsheets/d/1ohNcAfhtMiCl5ErlSf4IoAdy4r8G-HeKDf4i8VePnGM/edit#gid=0)

Does anyone have any idea?

poirqc
15-08-2015, 13:32
Certainly been an interesting read, and doing the test backs up what I've been feeling. Very handy though to pinpoint DZR and where scoop knee should be and how much reduction to apply.

215142

This is what came out for my test.

I used it and applied DZR of 0.06, using the 4:1, I put in DZF of 0.015, put the knee in at about 0.42 (0.46 actually due slider difficulties), and reduction of 0.44 or thereabouts (due sliders again).

I must say I'm liking the outcome, with all other sliders at default.

I should point in the first post that high Scoops Reduction values should be avoided. For now this tool is good to point FFB deadzone of a wheel.

PTG Ty1er Ward
15-08-2015, 14:04
@poirqc- do you think I should make a separate thread/post with a step by step walk through for DZR/DZF only? You don't need the graph for that, you can just look at the numbers. I think it can potentially benefit a lot of people and by simplifying/separating it I think may be less intimidating for people to try

poirqc
15-08-2015, 15:03
@poirqc- do you think I should make a separate thread/post with a step by step walk through for DZR/DZF only? You don't need the graph for that, you can just look at the numbers. I think it can potentially benefit a lot of people and by simplifying/separating it I think may be less intimidating for people to try

Well, imho, we should keep the information gathered at the same place. Before i decided to throw myself and start this thread, i had discussion that where scattered across the forum. I wouldn't even be able to find it again, sadly.

Id rather rename the thread, and separate the data in it instead. Or you could start a thread for now, since it would be easier for you to handle the updates. The 2 thread could be merge at some point when conclusions are found. I'm just trowing ideas here, i'm not a forum guru. :)

Even if i'm the "owner" of this thread, moderators could still edit the 1# post in case i'm not there to update it. It's what happened in the ps4 G29 thread.

I agree with you that there's alot more going on here than just linearity research. If there's enough data here, this(those) thread(s) could get stickied.

On a side note, Per wheel movement can make a world of a difference in the driving experience, this should get its thread of his own :D

konnos
15-08-2015, 23:01
@STEELJOCKEY Shouldn't your DRR be at 0.11 according to this chart? It's the point where the forces begin to appear.

STEELJOCKEY
16-08-2015, 01:04
I'm not going to hijac this thread and let it wander off in different tangents as it has been very informative so far, and I look forward to more research and understanding.

But to answer your question, on the Xbox the next choice would have been 0.12, and looking at the figures in the columns, the turning point was 0.07, trying to squeeze up the DZR from a wider point starts to make the wheel a bit notchy, and I've got to add too much DZF to smooth it back out.

poirqc
16-08-2015, 03:54
I'm not going to hijac this thread and let it wander off in different tangents as it has been very informative so far, and I look forward to more research and understanding.

But to answer your question, on the Xbox the next choice would have been 0.12, and looking at the figures in the columns, the turning point was 0.07, trying to squeeze up the DZR from a wider point starts to make the wheel a bit notchy, and I've got to add too much DZF to smooth it back out.

Imho, DRF is fairly easy to set. It should always be set in tandem with DRR. At the same time, it's fairly small if you compare it to DRR. As you move the DRR value, it can indeed get notchy. If you want to get on the low side for DRF, You can drive in a straight line, while doing small quick turn, left and right. If you don't have enought DRF, you'll be able to see the notches. As you move DRF up, the signal will look like an almost perfect sine wave. When you reach that, there's no need to put more DRF.

For what's it's worth, i reported what happened with different Deadzone Removal values (http://forum.projectcarsgame.com/showthread.php?35635-G29-Discussions-and-Support&p=1064827&viewfull=1#post1064827) - Forget about my conclusion, i would need to read the graph precisely, witch i didn't do.

I wish you get 3.0 quickly, it will enable the minutes adjustement for you too.

Cheers.

SeeNoWeevil
16-08-2015, 10:08
It would be really good if the original post could contain an introduction and basic guide. Coming in fresh I honestly couldn't even work out what the discussion was about.

STEELJOCKEY
16-08-2015, 10:54
Imho, DRF is fairly easy to set. It should always be set in tandem with DRR. At the same time, it's fairly small if you compare it to DRR. As you move the DRR value, it can indeed get notchy. If you want to get on the low side for DRF, You can drive in a straight line, while doing small quick turn, left and right. If you don't have enought DRF, you'll be able to see the notches. As you move DRF up, the signal will look like an almost perfect sine wave. When you reach that, there's no need to put more DRF.

For what's it's worth, i reported what happened with different Deadzone Removal values (http://forum.projectcarsgame.com/showthread.php?35635-G29-Discussions-and-Support&p=1064827&viewfull=1#post1064827) - Forget about my conclusion, i would need to read the graph precisely, witch i didn't do.

I wish you get 3.0 quickly, it will enable the minutes adjustement for you too.

Cheers.

I had my own blog thread a while ago in the xbox area, trying to work this stuff out, and certainly discovered the notchy bit and how to smooth it out.

Yeah, I'm hoping for 3.0 soon, but going on previous updates, if we xbox owners see it before the end of the month we will be jumping out of our skin with shock and fainting with excitement.

poirqc
16-08-2015, 12:09
It would be really good if the original post could contain an introduction and basic guide. Coming in fresh I honestly couldn't even work out what the discussion was about.

It's the plan indeed.

I actually got couple infos from your post Steeljockey :)

STEELJOCKEY
16-08-2015, 12:40
Did you? Which bit of my ramble made sense?

poirqc
16-08-2015, 14:16
Did you? Which bit of my ramble made sense?

Well, the discussion as a whole, the hand drawings, the questioning with Schnizz58. Everyone's input is valuable. I'm gathering the puzzle pieces here and there.

Back on topic, I'll try to update the first post soon and i start by asking conclusions first here.

The first important test is finding the FFB deadzone of a wheel. This is achieved with the Step Log 2(linear force test). The results from the test are tied to the drivers setting of a wheel. Someone should put DRR at the input value where the output starts moving. After that, DRF is set between a fourth to a half of DRR.

Scoops could be on the side of a user preference setting. However, a high reduction value should be avoided. Adding some reduction will influence the DRR needed.

I'd like to change the topic name, but i'm out of ideas.

Since english isn't my first tongue, if anyone have a better way of saying it, i'm all ears!

Cheers.

STEELJOCKEY
16-08-2015, 19:52
It's been a while since I've done a macro, but from memory it should be something as simple as:-

DZR=if output at row 1>0,use output number, if output at row 2>0, use output number, etc.
DRF=DRZ/4

You would probably need maybe a dozen what ifs to cover enough rows in case a wheel has a big deadzone in the software.

If the turning point you are looking for is actually the next row because the first one is still too flat, you could tell the macro to use the output number from next row, ie if output at row 1>0, use output number from row 2

Diluvian
16-08-2015, 21:33
215296

I'm still not sure, is this only a graph for us or are we able to compensate the non-linearity with a tool?

FYI my FFBOffset is 1% (minForce test)

poirqc
16-08-2015, 23:57
215296

I'm still not sure, is this only a graph for us or are we able to compensate the non-linearity with a tool?

FYI my FFBOffset is 1% (minForce test)

Wheel, it seems the DFP got a really low FFB deadzone. It's nice. If you don't use any Scoop, you could set the DRR/DRF to 0.01 and 0.005. If you use Scoops, you'll need to move the DRR/DRF values because adding Scoops will lower some input values. Some value may not be generated by the DFP because of that.

With that said, you have such a low deadzone that you may not need to move the DRR/DRF much.

Here's my updated sheet (https://docs.google.com/spreadsheets/d/1ohNcAfhtMiCl5ErlSf4IoAdy4r8G-HeKDf4i8VePnGM/edit#gid=0). You'll have to manually calculate your Scoops tho. You should be able to save it from there.

STEELJOCKEY
17-08-2015, 00:26
I've been having a think about what I've been feeling with the wheel, after applying DZR/F and scoop.

When I first looked at my graph and applied adjustments to my scoop settings (as large as they were), I was getting an extremely linear feel. Did my testing around Brno with the GT3 SLS, and noticed particularly in turn 1 that force would increase smoothly as I wound on the wheel. Felt perfect.

After applying DZR/F, the FFB seemed to go back to closer to default, as in without any scoop. Certainly not as bad.

I'm wondering if, when DZR/F is applied, it is happening first, skewing the whole curve to the left, and then scoop is being applied at the wrong spot.

Unfortunately I'm out of the 'office' for a few days, to retest this. I'd like to recheck the scoop settings without DZR/F, and see if it gets back that proper linear feel. If this is so, then after applying DZR/F I'd need to replot the scoop point and reduction value.

poirqc
17-08-2015, 01:01
I've been having a think about what I've been feeling with the wheel, after applying DZR/F and scoop.

When I first looked at my graph and applied adjustments to my scoop settings (as large as they were), I was getting an extremely linear feel. Did my testing around Brno with the GT3 SLS, and noticed particularly in turn 1 that force would increase smoothly as I wound on the wheel. Felt perfect.

After applying DZR/F, the FFB seemed to go back to closer to default, as in without any scoop. Certainly not as bad.

I'm wondering if, when DZR/F is applied, it is happening first, skewing the whole curve to the left, and then scoop is being applied at the wrong spot.

Unfortunately I'm out of the 'office' for a few days, to retest this. I'd like to recheck the scoop settings without DZR/F, and see if it gets back that proper linear feel. If this is so, then after applying DZR/F I'd need to replot the scoop point and reduction value.

DRR/DRF is applied after Scoops. However, i can't tell for sure how it affect the signal. I understand the Falloff but i'm not sure about the range. I should take some screen shot to be sure.

Diluvian
17-08-2015, 18:24
Stunning work poirqc!

Now I got how to compensate it in pcars.. :D

Here is my DFP setting and compensation (little bit tricky to calculate the scoop-curve with excel/google sheets) : DFP google sheet (https://docs.google.com/spreadsheets/d/1eRCcAHc3lCF1BUy2vgmjxT7gl-74dsdLOZ10fJLMZMM/edit?usp=sharing)

Now lets try it out and test what happens :D..

poirqc
18-08-2015, 02:30
Stunning work poirqc!

Now I got how to compensate it in pcars.. :D

Here is my DFP setting and compensation (little bit tricky to calculate the scoop-curve with excel/google sheets) : DFP google sheet (https://docs.google.com/spreadsheets/d/1eRCcAHc3lCF1BUy2vgmjxT7gl-74dsdLOZ10fJLMZMM/edit?usp=sharing)

Now lets try it out and test what happens :D..

I'm merely writing things, don't forget Tyler, Skoader, Shizz, dw123, VictorySV, etc(Sorry, it's late, can't search, will do)

First post updated.

Edit - BTW, how did you open the copy document to sheets, while staying as read only? I'd like to share mine this way.

Edit #2 - i'm not sure about your Scoop Calculation, the negative part makes me think. I'll have a look if i get the time.

Thanks

Diluvian
18-08-2015, 10:58
Sry, thanks to everyone else who helps in that project :)

You can simply copy the sheet and edit the copied version.

The negative part of the scoop is cause my reduction is set so high. Seems fine and makes sense in my eyes. If you find an error pls tell me :)

poirqc
18-08-2015, 11:03
You can simply copy the sheet and edit the copied version.


This option isn't available on my sheet, it's greyed out.

On yours, i was able to copy it as a google sheet, on mine, i can only download it in various format.

Diluvian
18-08-2015, 11:12
I don't know why you cannot copy it :D ..

I only gave you the link to "read" my sheet but you seem to be able to copy it. A little bit irritating.

eeeh what is that "K22" value on top right corner?

poirqc
18-08-2015, 11:20
I don't know why you cannot copy it :D ..

I only gave you the link to "read" my sheet but you seem to be able to copy it. A little bit irritating.

eeeh what is that "K22" value on top right corner?

Your document is read only to me, on the blue rolling menu. However, it's in the file menu i can copy it.

Your scoops value seems to be good. It gives you a DRR of 2%.


The idea around K22 is that id like to find a way to have and automatic way in telling where to start the Steep scoop fonction. I almost got something, but it's not ready yet.

Could i integrate your reduction/expension column in the sheet? I wanted to do something similar but yours is pretty neat.

Thanks.

Diluvian
18-08-2015, 11:27
Yes I copied it the same way..

Yep the automation of calculating the steep scoop function is a bit weird .. You can copy everything you want :) ..

[edit] wtf where is my file .. the linked sheet is wrong.

[edit2] lol now I got it :D .. https://docs.google.com/spreadsheets/d/1eRCcAHc3lCF1BUy2vgmjxT7gl-74dsdLOZ10fJLMZMM/edit?usp=sharing (fully automated) (use it as you want it, copy what you want if you like)

b_akerlite
18-08-2015, 12:27
nice work everyone!

does pcars have a setting in game that amplify output signals within the deadzone range?

my poor wheel has many hours on it and the deadzone is 24 which is alot., too much

Diluvian
18-08-2015, 12:29
The whole thread is about Deadzone removal and linearisation of your wheels output.

You can set the Deadzone Removal Range ingame to your percentage (24% = 0.24).. here more in detail: http://forum.projectcarsgame.com/showthread.php?22122-G25-G27-SETUP-GUIDE

b_akerlite
18-08-2015, 12:50
yeah i know the thread is all about that but thought there might be more to it or at least to how the game implements the drr or more settings to dial in.

i think these linearisation graphs are obviously prefect in respect to wheechecker - but for them to be accurate with the game don't we need to know more about how the game implements drr?

for example;

does the game look at your drr setting (In my case 0.24) and boost all ffb signals up to 24 by x ammount? what is this x ammount? the reason i am asking is that surely how the drr is implemented will skew the data in your graphs that is less than your drr setting?

poirqc
18-08-2015, 13:01
yeah i know the thread is all about that but thought there might be more to it or at least to how the game implements the drr or more settings to dial in.

i think these linearisation graphs are obviously prefect in respect to wheechecker - but for them to be accurate with the game don't we need to know more about how the game implements drr?

for example;

does the game look at your drr setting (In my case 0.24) and boost all ffb signals up to 24 by x ammount? what is this x ammount? the reason i am asking is that surely how the drr is implemented will skew the data in your graphs that is less than your drr setting?

I'd like to ask or do test about it at some points, i need to invent time! :D

Diluvian
18-08-2015, 13:52
yeah i know the thread is all about that but thought there might be more to it or at least to how the game implements the drr or more settings to dial in.

i think these linearisation graphs are obviously prefect in respect to wheechecker - but for them to be accurate with the game don't we need to know more about how the game implements drr?

for example;

does the game look at your drr setting (In my case 0.24) and boost all ffb signals up to 24 by x ammount? what is this x ammount? the reason i am asking is that surely how the drr is implemented will skew the data in your graphs that is less than your drr setting?

yep you're right, that would be nice to know in detail. I think they simply "stretch" the curve and kill the deadzone. So the most little signal will start at the point where the wheel starts to react (after 24% in your case). I think you loose resolution the more drr you set.

Diluvian
18-08-2015, 15:43
I added DRR and GAIN to the sheet. This only changes the "ideal linear curve" with the DRR reduction. The scooped output should follow that orange line as good as possible. I think this is how it is handled ingame, too.
If your wheel has a resolution/steps of 100 from 0% to 100% force and you set a DRR of 20% your wheel looses 20 steps and has left a resolution/steps of 80 from 0% to 100% force.

Link to sheet: https://docs.google.com/spreadsheets/d/1eRCcAHc3lCF1BUy2vgmjxT7gl-74dsdLOZ10fJLMZMM/edit?usp=sharing

Changelog:

- Added calculation for DRR to the linear curve
- Changed OUTPUT calculation from: =G2/G51*100 to =G2/G$52*100 (G52 is the last value!)
- ADDED GAIN Setting (simple gain, hopefully this is handled same ingame)

@poirqc and others, you can always use it for your official documents. If you find something which is wrong, pls tell me, too :)

ashwath84
18-08-2015, 16:00
Hi Poirqc,

Thanks a lot for all the effort on DRR/Scoop Calculations. I managed to create one for all the Logitech DFGT users. Hope it helps them (of course they need to insert their own values after running the wheel check test, but this should be a starter)

I have my Deadzone removal at 0.02 and fall off at 0.01 (almost nill i.e.).

I've entered the scoop knee as 50 and reduction as 25 and this gives me an almost perfect response when compared to the original response (without scoop).

Yet to try it out in game though :)

Link:
https://docs.google.com/spreadsheets/d/1kXLcvkRS6hcqYq37mf-SXRTK4pU7mOoTdjdNfKjp_1U/edit?usp=sharing

poirqc
18-08-2015, 16:57
Hi Poirqc,

Thanks a lot for all the effort on DRR/Scoop Calculations. I managed to create one for all the Logitech DFGT users. Hope it helps them (of course they need to insert their own values after running the wheel check test, but this should be a starter)

I have my Deadzone removal at 0.02 and fall off at 0.01 (almost nill i.e.).

I've entered the scoop knee as 50 and reduction as 25 and this gives me an almost perfect response when compared to the original response (without scoop).

Yet to try it out in game though :)

Link:
https://docs.google.com/spreadsheets/d/1kXLcvkRS6hcqYq37mf-SXRTK4pU7mOoTdjdNfKjp_1U/edit?usp=sharing

Before appliying Scoops, you have a DRR of 0.02. After appliying Scoops, it gives you a DRR of 0.04. You start to have output when the input reach 4%(0.04)

Thanks to you for taking the time to post your results.

Schnizz58
18-08-2015, 18:32
I added DRR and GAIN to the sheet.
The problem is that we don't know how much additional gain is added by the deadzone removal.

poirqc
18-08-2015, 18:37
The problem is that we don't know how much additional gain is added by the deadzone removal.

You do have a point, the Gain slidder isn't in the official guide.

On a side note, did wrote some wrong things in the first post?

Diluvian
18-08-2015, 18:41
The problem is that we don't know how much additional gain is added by the deadzone removal.

Yep you're right - this is still work in progress. The general gain should work like that. DRR Gain is something different which we don't know how it is handled, I don't know what the DRF does,too.

Again did some little updates: https://docs.google.com/spreadsheets/d/1eRCcAHc3lCF1BUy2vgmjxT7gl-74dsdLOZ10fJLMZMM/edit?usp=sharing

Added a new Graph which "cuts off" the amount of DRR you set up. There you can see the final result basend on my logic (DRR via cutoff, general gain settings, scoop settings)

poirqc
18-08-2015, 18:57
Yep you're right - this is still work in progress. The general gain should work like that. DRR Gain is something different which we don't know how it is handled, I don't know what the DRF does,too.

As we discussed, i would still need to precisely mesure this (http://forum.projectcarsgame.com/showthread.php?35635-G29-Discussions-and-Support&p=1064827&viewfull=1#post1064827)

Schnizz58
18-08-2015, 20:18
Yep you're right - this is still work in progress. The general gain should work like that. DRR Gain is something different which we don't know how it is handled, I don't know what the DRF does,too.
DRF reduces gain (linearly I assume) on the edges of the deadzone reduction range. The larger the value the more gradual the dropoff in gain.

Diluvian
18-08-2015, 22:05
You mean something like that?

215489

(Sorry for bad paint-skills)

maybe the use lower input-signals (blue line) lower than the DRR-Value? (0% DRF means, they use DRR als the lowest possible input value and 50% DRF means they use input signals inside the "deadzone"?). The orange dropoff-lines seems unlogic to me.

I tested that behaviour with 99% DRR and high DRF .. the wheel uses lower forces which means they use the input values lower than the DRR has defined (my blue lines at the image). If DRF = 0 the wheel is only able to respond with max force (square-graph on telemetry).

Schnizz58
18-08-2015, 22:22
No, the DRF would be on the other side of the DRR point. It applies outside the deadzone range, which in your example above is 0.25. So the falloff zone would start at 0.25 and go up from there, depending on how much falloff you specified.

The response curve of deadzone removal is like a plateau with the width of the plateau defined by DRR and the slope of the sides by DRF. At least that's how I understand it. I'll try to draw you a picture but I don't have time right now.

Diluvian
18-08-2015, 22:24
But that would mean if you set 1.0 DRR and 0.1 DRF (maximum) the wheel would not react smoother than with 0.0 DRF. But that isn't how the wheel in pcars reacts, it's nearly completely smooth (like it could use the complete input-range). That would mean that your "plateu" is the rest of the active area (75%). The more I think about it the more I think it's the blue line (but it doesn't feel linear at all?). This setting is propably meant to smooth out the wheel if the wheel force at the DRR-Point isn't nearly zero, so you can give it the ability to use values lower than DRR as the minimum-input. .. hm

Schnizz58
18-08-2015, 22:44
But that would mean if you set 0.99 DRR and 0.1 DRF (maximum) the wheel would not react smoother than with 0.0 DRF. But that isn't how the wheel in pcars reacts. That would mean that your "plateu" is the rest of the active area (75%). The more I think about it the more I think it's the blue line. This setting is propably meant to smooth out the wheel if the wheel force at the DRR-Point isn't nearly zero, so you can give it the ability to use values lower than DRR as the minimum-input. .. hm
I'm not sure what your point is. The purpose of deadzone removal is to boost the gain at low forces (and therefore low steering angles). So it has no effect above a certain force.

Diluvian
18-08-2015, 22:47
I'm currently struggling with showing what I mean :D .. give me some minutes, maybe I get it on paper. But to answer your question: Take 5min. and set pcars DRR to 1 and DRF to 0 .. you expect a square response from the wheel (the wheel reacts very hard). Now set DRF to max and you see the wheel is able to use lower forces which means that the software sends input signals lower than the DRR is set.

Schnizz58
18-08-2015, 22:57
you expect a square response from the wheel (the wheel reacts very hard).
No I don't.

Diluvian
18-08-2015, 23:03
What then do you expect, you remove 100% of the possible input values, only the maximum you can put in as minimum value.

Heres my scetch:

215491

I don't know if it's understandable what I have scetched there.. the point is that the green line is only with DRR (the right graph shows the "new" input-output-response). The red line is with DRF. The software is able to use values lower than DRR. It seems we understand the DRR/DRF thing completely different :p

Schnizz58
18-08-2015, 23:06
I wouldn't expect to remove anything. All it does is boost the gain at low forces.

Diluvian
18-08-2015, 23:18
Look at the right graph at my drawing, there you have the "boost" you want at low values. As you can see, an ideal response curve with DRR would result in high forces at low input values. Lets have a look at a response which isn't ideal:

215492

Thats how I understand DRR

Isn't some DEV here? :D

PTG Ty1er Ward
19-08-2015, 00:57
I'm not sure what your point is. The purpose of deadzone removal is to boost the gain at low forces (and therefore low steering angles). So it has no effect above a certain force.

This is where the confusion lies. I tend to agree with Diluvian (that was my initial thought on how it worked). DZR is an FFB setting not a steering input setting (that's a separate DZ removal setting). Why it ended up being called "Deadzone Removal" I don't know- the original name for it was "Tighten Center". I don't think it has anything to do with with the the steering angle, but that would be the most noticeable (and beneficial) improvement. Your basically telling the FFB engine- "boost forces below (25 in his example) by X amount, or this fraction of X or this algorithm of X or whatever it is".

IMO it increases the forces only BELOW the DZR setting (like a an opposite mini version of scoop). The "falloff" is the percentage increase towards 0 (similar the to blue lines in his drawing).

PTG Ty1er Ward
19-08-2015, 01:26
I've been running back and forth between Linear input/output with no ffb corrections default TMCP settings to feel and see the difference. I've held off posting with my observations because there are a few more things I want to try, but I will say this for now- I can't quite say for certain exactly what the FFB in telemetry represents. You can have little to no feeling in the wheel and the FFB in telemetry is clipping hard and going crazy, and the opposite can also be true, the FFB meter is barely moving and the wheel is heavy with earth shattering FFB.

Schnizz58
19-08-2015, 02:38
I don't know how to explain it any clearer. Deadzone removal applies additional gain at low forces to increase those forces so that you can feel them better. If you guys don't want to believe that then fine.

If it worked the way y'all are describing, there would be a huge notch at center.

poirqc
19-08-2015, 03:32
I don't know how to explain it any clearer. Deadzone removal applies additional gain at low forces to increase those forces so that you can feel them better. If you guys don't want to believe that then fine.

If it worked the way y'all are describing, there would be a huge notch at center.

The question isn't IF it increases low forces, HOW it increases them.


TightenCenterRange (​Deadzone Removal Range​)
This is the input force below which the output force is increased to remove a deadzone. Put
more simply, this is the size of the deadzone you are trying to remove.
TightenCenterFalloff (​Deadzone Removal Falloff​)
This controls how softly (higher values are softer) the output force approaches zero force as
the input force goes below TightenCenterRange.


Ok, let try to analyse that.

How about: DRR is just the width of the FFB Deadzone. Without any DRF, you have a "plateau", and this give a square wave in the telemetry hud. DRF is the slope of the deadzone.
215503

Does it make sense?

If it does, should we calculate the DRR differently since if you try to generate some falloff signal where the wheel can't output anything, it would serve no purpose?

PTG Ty1er Ward
19-08-2015, 04:40
I don't know how to explain it any clearer. Deadzone removal applies additional gain at low forces to increase those forces so that you can feel them better. If you guys don't want to believe that then fine.

If it worked the way y'all are describing, there would be a huge notch at center.

Ok now I'm confused lol. You just described what I was saying but you seem to think I'm saying the opposite :)

Sidenote- set your "wheel position smoothing" to 0 (instead of the 0.04 I'm sure your at) and feel the notchiness :) with the dzr and dzf set correctly you can run a 0.00 position smothing (which is a mix of current and past forces). Except on Xbox you can't because the default steering gain is at 3 so it's super notchy (until you can tweak it when 3.0 is released). I'm telling you guys, the more I learn the more I realize I'm barely scratching the surface.

PTG Ty1er Ward
19-08-2015, 04:44
The question isn't IF it increases low forces, HOW it increases them.



Ok, let try to analyse that.

How about: DRR is just the width of the FFB Deadzone. Without any DRF, you have a "plateau", and this give a square wave in the telemetry hud. DRF is the slope of the deadzone.
215503

Does it make sense?

If it does, should we calculate the DRR differently since if you try to generate some falloff signal where the wheel can't output anything, it would serve no purpose?

You were right in the beginning in your other thread about DZR/DZF. Just pick a combo that feels right to you. The most important thing is per movement/movement squared and Smoothing (the goal should be no nochtyness at 0 for smoothing). I'm running 0.01 and it's flawless. That might be the best I can do with the TX but I'll still try a few other things.

ashwath84
19-08-2015, 08:27
Before appliying Scoops, you have a DRR of 0.02. After appliying Scoops, it gives you a DRR of 0.04. You start to have output when the input reach 4%(0.04)

Thanks to you for taking the time to post your results.

Thanks mate :)

Diluvian
19-08-2015, 10:45
I really took the time and wrote down some text but I deleted it .. I think it's time to get a clear answer from some dev to be able to continue.

poirqc
19-08-2015, 11:06
You were right in the beginning in your other thread about DZR/DZF. Just pick a combo that feels right to you. The most important thing is per movement/movement squared and Smoothing (the goal should be no nochtyness at 0 for smoothing). I'm running 0.01 and it's flawless. That might be the best I can do with the TX but I'll still try a few other things.

Witch one? I'm losing track of them. :)

215523

Would it be something like that where we would need to have a cross over and push DRR a bit further after the wheel can output a force by itself? That way, the whole range of forces generated by pCars could be output by the wheel?

I think Per Wheel Movement is about "offsetting" the input insteand of "offsetting" output like Deadzone Removals and Scoops do.

What do you think.

inthebagbud
19-08-2015, 11:25
I really took the time and wrote down some text but I deleted it .. I think it's time to get a clear answer from some dev to be able to continue.

The thread is very interesting but I had to bow out after initial posts as until you are informed by Devs what all the settings actually mean and how the initial game setups affect individual wheels you could be going round in circles - especially where the consoles are concerned

The in game defaults must have been modeled on some wheel, why they cannot say which is beyond me as this would give you a baseline to work from

I applaud your efforts and knowledge and I appreciate the PC version can be modeled differently as you can use the wheels control panel , but there should be some input from SMS even if its privately to tell you whether you are on the right track with your theories