Thrustmaster t300 Recommended FFB MPs each car

Firstly, nice illustrations again Paul.

And I'm glad you drew them because i think i now know where our misunderstandings lie. My belief of how the STS function works seems to actually agree with your (verbal and illustrated) perception of how the STS affects the FFB output in your experience. Let me explain.

Firstly, i've used the following changes in FFB output response (which is the final output sent to your ffb wheel = y-axis) to the raw FFB output (that just came out of the physics engine = x-axis) to represent the general effects of linear, low and high STS values on the FFB output signal (based on my hypothesis).



and here's how the FFB output over time of some arbitrary racing scenario is affected by different STS values. (Blue being the default ffb output everyone would feel/experience if they did not change the default STS value in the controller.json file).



Notice that it's similar to your illustration of my interpretation of how STS works in terms of how the peaks are sharper/narrower for lower STS values and blunter/wider with higher STS values. However the similarities end where the height of each peak is concerned. As you can see the height of the peaks and troughs (provided they are not 0% or 100% of max torque outputs) are different for each STS value whereas you drew them to be the same for each STS value. It's only when the FFB output is at 0% or 100% of max ffb torque that each STS value has the same ffb output height/amount.

Now, i can't help but get the impression from your first illustration that low STS in your experience has the effect of amplifying/exaggerating the difference between different ffb outputs. If you look back at my graph, whilst perhaps not immediately apparent, you can see that this is also happening to the low STS. For example, compare the relative difference in height between a peak and trough for the high and low STS FFB outputs. For example, the peak at time 2.5 and the trough at time 4. For the high STS, the difference is 63/40 = 1.58x and for the low STS, the difference is much greater at 16/2.5 = 6.4x. Again for the peak at time 15 and trough at time 10. High STS difference is 89/63 = 1.42x but for the low STS the difference again is far greater at 64/16 = 4x. You can repeat and compare this analysis with any pair of peaks and troughs (provided a non-zero trough) and you'll find the percentage difference is much greater for the low STS than the high STS in every instance. And it's this greater relative difference in ffb output amounts that makes lower STS values (i believe) feel as you've described both verbally and in your illustration.

Apologies msportdan for going off topic in your thread. Hope you don't mind. Please let me know if you do.

 

Ok, so there is a problem with making such an analysis/comparison as always and in comes from the fact that the torque output from the motor acts as an input remember due to it being a conventional loop. So if for example the low STS decreases the ffb output from the motor (for everything below the max torque output relative to the raw ffb output = linear STS), then the resultant ffb steering wheel path/trajectory (caused by the ffb signals sent from the sim) will change and when that trajectory changes so does the ensuing/future ffb outputs from the sim. The end result is that in reality you would not have the exact same general ffb output pattern for each STS value because to change the strength of the ffb output (which STS does) would change the trajectory of the ffb steering wheel which in turn changes the future ffb output trajectory which in turn changes the perceived ffb response. So you can't really look at this in an isolated manner but if we momentarily assume that we can, then we compare and see how STS would change the ffb output for the same ffb output trajectory (if that makes sense).

Also, don't fixate on the curves in the second graph. They don't actually mean anything. They are just a consequence of having a non-linear relationship between the raw ffb signal and the final ffb output to the ffb wheel (as shown the first graph for different STS values). In the second graph, you are just looking at the torque signal sent to ffb wheel at different moments in time (x-axis). For example, assume that each instance along the x-axis represented a frame/moment-in-time, then at frame/time 5, the low STS ffb is simply telling rf2 to send an ffb signal/demand of 15% max motor torque instead of the default/linear STS of 40% of max motor torque (which is the same as raw ffb output) or the high STS of 63% max motor torque.

Working out the change in the raw ffb signal to the correct ffb output based on the STS value chosen is very simple. You simply perform a function on the raw ffb output to get the corresponding STS ffb output. Here's how it works.

For the linear STS the linear function y=x is used. For the low STS, i used a function of y=x^2. For the high STS, i used a function y=x^0.5. (these functions are what produced the ffb response curves in the first graph)

Before we plug in the raw ffb output, we have to convert it from some value between 0-100% to some value between 0-1. (E.g. if raw/linear STS ffb output was 40% = 0.4). Then for the low STS, the corresponding ffb output = 0.4^2 = 0.16 which equates to 16% max motor torque. For high STS, the corresponding ffb output = 0.4^0.5 = 0.63 which equates to 63% max motor torque. And ofc, the extremes of 0% and 100% result in the same ffb output to the ffb wheel in all 3 STS cases.

Hope that makes sense.

 

I'm really not talking about the exactness of the torque response of the t500 or assuming you think the t500 is giving you accurate feedback.

Not intentionally so but this happens to be partly the reason for this problem. I mean think about it. With the alternative loop, you wouldn't even be able to have an STS function (let alone need one). But that's besides the point, you made a point of "Your graph would make sense in a pre-defined scenario where the outcome is already known" which you were correct in saying and i was explaining why my analysis (that was in fact an extension of your own originally) doesn't make sense in reality yet if we momentarily assume it does then we could compare the effect of STS to one another (which is really the only way we can). But i don't want to go down that rabbit hole. The topic is how the STS works and not why we can't make a comparison of the effects to ffb output between different STS values in reality because we can glimpse it in theory (which is what our graphs/illustrations are doing).

You need to explain what your graph represents exactly then because it doesn't make sense in that case. I thought the x-axis was time and the y-axis torque output/demand from the motor. Is that correct?

Another thing, motors can change torque nearly instantaneously for every torque demand possible from the motor. For every ffb frame, the sim sends our ffb wheels some ffb output signal/demand that is unchanging for the duration of that frame. After that frame, the position of the steering wheel is returned to the sim, the physics engine computes the next restoring torque that the ffb wheel should output and the loop repeats. There isn't even a lot of time for the torque to more gradually build up over time because each frame in the case of rf2 is only 2.5ms. I don't even understand how the torque could be made to build up gradually. Your description of how STS is working does not seem to fit the capabilities of the technology being used.

 

Same here with a Fanatec. It's the way the cars are programmed. You're mostly wasting your time fiddling because to get the "bad" cars to feel better, you will have to ruin the "good" cars. Or, spend who knows how many hours fiddling with car-specific controller settings.

Does the new BT20 feel awesome right out of the box (all settings at default and 1.0 multiplier)? Sure does for me. The Megane feels great. The Clio feels great. The ASR F1 cars feel beyond great. The new Palatov D4 under development feels great. How then, at the exact same settings, do several other cars including the worst offender, the F2, feel like crap? I will just wait (im)patiently until ISI gets around to fixing them.

I know that there are several different FFB approaches that have been tried. Supposedly the updated Brabham has the latest/best. Well, bring it on for the other cars. There is just no way that arcane .JSON settings and fiddling to such an extent is/should be necessary when 100% default settings produces the most sublime FFB of any consumer-level sim yet on a whole range of vehicles. The problem is with the cars, not your wheel or settings.

That being said, it is very difficult to get widely varying hardware to all feel good. But so far, I haven't heard anything but good feedback on the new Brabham. Can you confirm? If so, then ISI may have finally hit the sweet spot and can update the other cars with whatever this latest FFB approach is for the Brabham.

 

I will reply in full tomorrow but a few issues I have with what you said is that the graph is torque along the x-axis and time/distance along the y-axis. I think you meant the other way round but still you can't have a time/distance axis. These are two different physical quantities that cannot be bound to the same axis.

Also, torque cannot represent speed because torque is an acceleration. You lowered the sts to reduce the acceleration rate (i.e. ffb torque output) of the steering wheel that was accelerating into the turn and thus into the slide too fast. That fits my hypothesis of how sts works perfectly and as shown by my first graph. The low sts ffb output curve generates less torque output from the motor than the linear sts curve for any values greater than 0% and less than 100%. So for example if the restoring torque were 40% and you let go of the steering wheel and found it turning (technically accelerating) too fast then using the low sts curve in my graph would instead produce a 16% torque that would produce a smaller acceleration that would be perceived I guess as a slower average turning speed into the slide.

Also, I was never under the illusion that sts and ffb multi are the same and/or had the same function.

 

Ok so it's supposed to represent distance axis? If so, I have no idea what a torque vs distance graph is supposed to represent.

 

Yes, it's very convoluted so let me try and unpack it.

Let's take the FFB multiplier out of the equation. Let's assume that for some specific car we setup the ffb multiplier to a value that maximises use of our ffb motors torque output range whilst not hitting any ffb clipping issues whilst driving at the limit. That should take care of the ffb multiplier issue and remove it from the following discussion since changing STS from here on out will not affect the ffb multiplier.

Next, let's look at your previous example which i liked of a car traveling at a constant speed of 20mph, at a constant radius of 100ft. Let's say that the with the default/linear STS the ffb output signal sent from the sim to the ffb wheel was 50% of maximum ffb motor torque. So holding the steering wheel at the required angle to maintain this requires you to apply a constant driver input torque equal to 50% of the maximum motor torque. Now let's also say that you let go of the ffb steering wheel, this 50% of maximum ffb motor torque (no longer resisted by you) causes the ffb steering wheel to accelerate out of the turn too fast and you think for it to behave realistically you want it to accelerate out of the turn at half that rate. Let's use my graph to see how the low STS curve achieves this:



50% torque along the x-axis (the raw ffb output) gives us 25% torque output along the y-axis (the final ffb output signal to be sent to the ffb wheel) for the low STS response curve.

So we repeat the same experiment of a car traveling at a constant speed of 20mph, at a constant radius of 100ft. So holding the steering wheel at the required angle to maintain this requires you to apply a constant driver input torque equal to 25% of the maximum motor torque (half of that before the STS was changed from default/linear to low). Next you suddenly let go of the ffb steering wheel, since your ffb wheel is outputting only half the torque as before, the steering wheel accelerates out of the turn at (roughly*) half the rate it did before and the steering wheel response now feels realistic to you.

Lastly, i think you mentioned that if the ffb motor has 1:1 torque output capability then you wouldn't have this issue but in fact this issue would become exacerbated. The reason is simply that with the same torque outputs, since the ffb wheels inertia is a ton less (e.g. 22x less in my approximation) than in a real car, the same torque output will accelerate the ffb steering wheel (22x) faster. Which only further causes the very problem you mentioned of the steering wheel turning itself too quickly when you let go of it. Again, the reason for that comes back to the flaw in the conventional loop method.

note: * It would not accelerate out the turn at exactly half the rate because there is also inertia of the ffb wheel at play, continuous but changing torque outputs as the steering wheel angle changes from being accelerated out of the turn, etc, all affecting the final average-acceleration/average-speed of the steering wheel turning itself when you let go of it. But i hope you can see that the point remains the same.

 

The graph is a way to visually demonstrate/illustrate to another my hypothesis of how sts works and your observations of how the sts works in rf2 seem to agree with it. I'm not trying to force your observations of how sts works to agree with my hypothesis, they already do.

Your graph was not helpful either Paul. Clearly we both do not understand each others graph. I already gave an explanation without the graph but you didn't understand it so I brought the graph in to help but clearly seems to have done the opposite. I don't think I could demonstrate/explain it any other way.

 

My two penneth: DrR1pper is absolutely correct in his interpretation of how the steering torque sensitivity works.

His simple graph explains it perfectly, no need for any long winded explanation. It's a simple mathematical function applied to the FFB output from the sim (which ranges from -1.0 to +1.0). It works in exactly the same way as when we apply non-linear curves to the steering, throttle, brake & clutch inputs, for example. Dragging the sensitivity sliders in the controller settings UI shows us exactly what happens when we move the sensitivity from linear (100% in the UI or an STS of 1.0) to non-linear - we see a curve exactly like DrR1pper's graph.

That's really all there is to it.

 

When you say the differences in the peak torques when turning onto the flat bit of the corner is 70% less between STS 0 and 2, in which direction is this difference? E.g. STS 0 has peak torque that is 70% less than the peak torque of STS 2's or vice versa?

If STS 0 peak torque in the same scenario is less than STS 2's peak torque, then that agrees with the graph.

 

Ah ok, so your experiences agree with my graph except for the very max torque?

So if at STS 2, in a scenario where the ffb bar just clips into the red (implying 100% max ffb motor torque), STS 0 does not also just clip into the red if you repeat the exact same scenario?

 

I just realised a way to test if the ffb max torque percentage is the same for both STS 0 and 2 or not.

Repeat the oval experiment with sts 2 first and set the car specific ffb multiplier so that the ffb bar just clips into the red and holds it there for the duration of the corner. Then repeat the experiment with sts set to 0 and if the ffb bar is also holding in the red during the corner than it is working like my graph. If not then my graph is wrong.

 

Paul, regardless of the sts value chosen, the ffb bar represents a range of virtual steering torques that is to correspond to the range of percentage of max torques from our ffb wheels. So if for example when setting the car specific multi to some value that gave you a maximum virtual steering torque of 10Nm that corresponds to the 100% of max torque from your ffb wheel. Now regardless of whether you set the STS to 0 or 1 or 2, the 100% of max ffb motor torque will still correspond to 10Nm of torque at the virtual steering wheel.

Only the ffb multiplier that can change the relationship between maximum virtual steering wheel torque and the 100% of max torque from the ffb motor.

 

I'll be home in half and hour and I'll reply then.

 

Would it be helpful to use devmode and change the vehicle's max torque so a normal, reproducible force is approaching max torque and can be more easily tested?

 

Ok, i'm understanding what your saying now but i think it's wrong. Like i said before, the only function that can alter the ffb wheel's 100% torque to represent some maximum virtual steering wheel torque is the car-specific ffb multiplier and STS function has no dominion/say over this. All that the STS can do is alter the mathematical relationship between virtual steering wheel torques to ffb wheel's torque range (except for the 0% and 100% torques).

Look at this:



The top graph is example of how the FFB multi function changes the range of virtual steering wheel torques that corresponds to the range of ffb motor torque output (always represented as a "percentage of the motors max torque output" because the physical amount of torque output at an ffb wheel depends on the motor powering it ofc....so the same ffb signal of 50% of max motor torque may produce only 2Nm of torque on a t500 wheel whilst on a bodnar/AF wheel it might produce 7-8Nm of physical torque output.) There is no need for the sim to know what the ffb wheel is capable of outputting....the ffb wheel simply receives an ffb signal that is a percentage of it's max possible torque and the ffb motor simply outputs this percentage. So if the ffb wheel receives an FFB signal of "0.0" then it means set the motors current to zero and hence no torque output. If it receives an ffb signal of "1.0" then it means to the ffb motor set the current to it's maximum (thereby producing maximum torque). For any ffb signal between "0.0" and "1.0", the ffb motor scales the current supply accordingly. So if it receives and ffb signal of "0.5" then it sets the current to half it's max. (I think it's usually a controller inside the ffb wheel that will will interpret and scale the current supply according to the ffb signal received).

You can see that increasing/decreasing the ffb multi simply increases/decreases the maximum virtual steering wheel torque that corresponds to 100% of max ffb motor torque.

In the second graph, i took the case of the FFB multi = 1.0 which has a max virtual steering wheel torque of 10Nm. The y-axis represents the virtual steering wheel torques and the x-axis represents the corresponding ffb signal to be sent to your ffb wheel based on which STS curve you are using. So in my arbitrary example, if your using a low STS then for (example) a virtual steering wheel torque of 5Nm (=50% of the max virtual steering wheel torque range set by the ffb multiplier) corresponds to only 25% of the max ffb motor torque output. High STS corresponds to 75% of max ffb motor torque. And linear STS ofc corresponds to 50% max ffb motor torque.

Now if you increase or decrease the ffb multiplier, the shape of these STS curves in the second graph will remain exactly the same. The only difference will be along the y-axis where the max virtual steering wheel torque will either increase or decrease based on whether you've increased or decreased the ffb multiplier.

But in each STS case, if the physics engine is saying that the virtual steering wheel is experiencing the maximum virtual steering wheel torque (as determined by the ffb multiplier) then regardless of which STS curve you are using, the ffb wheel will output 100% of its max motor torque. This can be seen as the 3 STS curves converge at the top right of the second graph.

My point earlier was that we can test this and see if my hypothesis is correct by repeating the oval experiment. Firstly by setting the STS to 2 and the carefully calibrating the ffb multi so that when in a stable and repeatable turn, the ffb signal is only just clipping into the red. Then repeat the experiment with the exact same ffb multi but now with the STS at 0 and see if the ffb signal is also just clipping into the red or not in the same cornering scenario. If it is, then my hypothesis is correct. If it's not then it's incorrect.

I would do this test myself if i could but i simply don't have the hardware to test on. It's all packed away in storage.

 

I'm not disagreeing with you there. We just seem to have different interpretations as to what/how it's doing things differently.

I made some edits and new additions to my post whilst you were replying. I gave a reason why i cannot test it myself and a way in which we can test my hypothesis to know for certain if it's correct or not.

100% agreed and it's shown in my second graph.

I'm sorry paul but that doesn't make sense. How can it control how long a torque output lasts? This is all determined by the loop. The physics engine computes some virtual steering wheel torque. It sends it to the ffb wheel as some percentage of maximum ffb motor torque and the ffb wheel outputs this torque for the duration of 1 ffb frame (in the case of rf2's, for 2.5ms). After which the physics engine computes a new torque based on how the steering angle was changed by the player over the last 2.5ms. This new ffb torque is sent to the ffb wheel where it once again outputs a new torque for the next 2.5ms. Rinse and repeat.

Even if you are/were correct in saying that sts somehow controls how long the torque outputs last....i haven't heard or seen a reasonable explanation for how this can happen given what we know of the hardware and physics loop.

Are you sure that you're not just mistaking how long it lasts for how big the torque output on your ffb motor is? I mean, your screenshots seem to show that for the same virtual steering wheel torques, the ffb output bar (which represents the ffb output send to the ffb wheel) is simply lower for a lower STS setting. Which is consistent with my graph. The only thing remaining to be tested/confirmed is whether the max virtual steering wheel torque causes the same maximum ffb output signal for any STS value and i explained how one could test it in my previous post.

 

I really do not understand what you mean them. For starters, rotational speed isn't even a force.

I don't understand why you keep coming back to the ffb multiplier. I've not been saying that you should change it when comparing the affects of different STS values.

Huh? Look back at the second graph. All 3 sts curves converge in the top right hand corner. This is where the maximum virtual steering wheel torque = the maximum ffb motor torque output in all 3 cases.

Changing the FFB multi streches/scales the virtual steering wheel torques to the ffb motor torque outputs regardless of the STS value.



See how the STS curves are characteristically the same regardless of the change in FFB multiplier (thus change in maximum virtual steering wheel torque to max ffb output)?

I don't understand what it confusing or conflicting about this.

 

Yes and that's fine. Do not take my pictorial representation of how differently the extreme opposite STS values can affect the ffb non-linearity. I mean, for all we know they could look like this when set to absolute low (0) and absolute high (2):



And with this one i've drawn, you can see for the same virtual steering wheel torque of 50%, the difference between the low and high STS ffb torque output is a whooping 15% and 85%. Which is pretty much the sort of difference in ffb torque output we can see in your rf2 screenshots between STS 0 vs STS 2. I simply do not know the function that rf2 uses to change the linearity of the ffb response but if i did i could predict for you the exact ffb output for any STS value.

Well, like i've been saying. We can remove all opinion and personal incredulity from the debate by simply running the test/experiment method i mentioned in my previous post. Could i be wrong? Yes. Could you be wrong? Yes. Could we both be wrong? Also yes. The only way to know for certain is to perform the test and review the evidence.

Yes, i know which is why despite my convictions that i'm right about this, I've been calling it a hypothesis all along. Something to be tested and confirmed true or false. So far your observations for everything but the max torque has agreed with my hypothesis. Whether it also agrees in the boundary condition (i.e. max torque scenario) is the last bit that needs testing to know whether the hypothesis is truly correct or not.

 

Oh, i must have missed that post. Which one was it?

So at STS 2 when in the corner the ffb bar was staying in the red the whole time but for STS 0 it was not? If so, what was the ffb bar holding at for STS 0 vs the constantly clipped ffb bar for STS 2?