Thrustmaster t300 Recommended FFB MPs each car

Steering torque is displayed at +- 10Nm. To scale it the same as FFB I need to use NominalMaxTorque (6.0Nm) instead.

Sent from my SM-G900F using Tapatalk

 

Think you mean 6.0 (modified from the 9.5, as you said). The STS_1 graph appears to show the FFB output (logically) being capped at 1.0 when the torque exceeds 6.0, though the examples are only fleeting so wouldn't call it 100% proof.

It's all good, must admit I wasn't expecting what I said to be a contentious issue and wasn't aware of your guide. I only checked one car and modified its nominal max value once, so not quite exhaustive testing... And I 'went quiet' because I was at work lol

 

Perhaps. It's just that you'd then have to present 3 graphs instead of just the 1 to explain/demonstrate the same thing. But, each to their own.

 

Oops, yeah I do mean 6.0.


Sent from my SM-G900F using Tapatalk

 

lol, even if contentious i'm happy you raised it. I'd much rather end up having to stand corrected that unknowingly continue believing in something that is wrong. Whilst I'm still not 100% convinced though (for reasons i've already explained), the evidence thus far does seem to agree with you.

Phew! I did get a little concerned i might have accidentally rubbed you the wrong way.

 

Sure and like i said all along, it was only a hypothesis regardless of how strongly i happened to believe it was correct.

I 1000% understand and respect that you need to see evidence before you are able to commit to a belief that some hypothesis is correct. Without that, it's just as likely to be wrong as it is to be right in your eyes (as it should be). Skepticism is a good thing.

Having said that though, my follow up questions were an attempt to get you to do some more specific/rigorous tests that i think would have shown you the complete evidence in support of my hypothesis before Ade was kind enough to post some of his own. But i know you were away from your pc for the weekend and therefore unable to, so it's a mute point.

Again, i would have chipped in with some of my own time spent experimenting but i simply don't have the time atm nor is my ffb wheel readily available. I almost went and picked it up from the storage unit today but i've just been too busy going back and forth to visit my nan in hospital for the last couple of weeks.

 

From what i'm being told, with the default ffb multi set to 1.0, if NominalMaxTorque = (for example) 9.5Nm then the range of virtual steering wheel/shaft torques that matches the range of ffb output (i.e. 0-100% which produces a different range of physical torques depending on the ffb wheel you have ofc) will be 0-9.5Nm for 0-100% ffb output signal. I.e. whenever 9.5Nm of torque occurs on the virtual steering shaft, the sim will send a 100% ffb signal to my ffb wheel (which for a t500 will result in around 5Nm, for a csw v2 6.5Nm, for an AFPro 13Nm, bodnar 16Nm, etc). If the virtual steering shaft is ever more than 9.5Nm, the sim will still send a 100% ffb signal (causing ffb clipping).

But I'm also being told that if i lower the ffb multiplier the NominalMaxTorque value does not increase. The reason i have trouble believing how this can be true is because that it seems diametrically opposed with my own extensive observations/experience of the ffb multiplier effect.

How exactly?

Because for most cars (about a year ago), the default ffb multiplier of 1.0 would cause ffb clipping within the driving limit/envelope (i.e. in the high speed corners the ffb would clip which is a bad thing ofc). The way i seemed to be able to fix it was by lowering the ffb multiplier, sometimes to as low as 0.5 for some cars. This had the observable effect of moving the ffb clipping region to outside the driving limit/envelope (i.e. ffb clipping no longer occurring in the high speed corners or just tittering on the edge of clipping when driving at the limit). Observable in TechAde's plugin and also very observable through your hands when driving at the limit. The torque output was no longer producing an unchanging max/100% torque output from the motor in the high speed corners but instead dynamically changing, providing detailed information about the car whilst doing so with torques that were very close to the max torque output possible by the ffb wheel.

Now for that to have worked, the only way i can see that being remotely possible is if the NominalMaxTorque was increased as a result of lowering the ffb multiplier. Why? Because the ffb clipping issue was caused by the fact that the virtual steering shaft was experiencing more than (for example) 9.5Nm of torque in the high speed corners (at ffb multi = 1.0). So surely the only way to avoid this would be to have the NominalMaxTorque raised just enough so that whatever the max virtual steering shaft torque actually was during the high speed corners, it was below the NominalMaxTorque which correlates with the 100% ffb signal. If the max virtual steering shaft torque the car/you experience when on track is above the NominalMaxTorque value set for the car, then you will encounter ffb clipping.

And so, I cannot see another possible explanation for how my observations of lowering the ffb multiplier manages to avoid the ffb clipping issue unless it raised the NominalMaxTorque.

Now, provided my observations were correct, is there another explanation that can also explain my observations that I'm simply not seeing?

 

From what I can see it works like this...

1. Steering shaft torque is calculated
2. FFB Multiplier is applied to calculated shaft torque
3. Multipied steering torque (Nm) is converted to a percentage by dividing torque by NominalMaxTorque (where it is capped to ±1.0). Sensitivity curve is applied here.
4. FFB filter is applied (rolling average of last X samples?)
5. FFB output sent to wheel is the output from the filter multiplied by Steering effects strength (from Controller.JSON). This converts the output into the range expected by DirectInput i.e. ± 10000.

From my original data we can see that the steering shaft torque isn't limited to NominalMaxTorque, it can go above it i.e. the clipping happens later in the chain.

 

Here's some data with smoothing at zero and FFB & Steering Shaft Torque displayed at the same scale.

STS=0
https://www.dropbox.com/s/jo6qg9n3g5z4ajw/STS_0.0_NoFilter.png?dl=0

STS=1
https://www.dropbox.com/s/7n8ayusj4lv0mhk/STS_1.0_NoFilter.png?dl=0

STS=2
https://www.dropbox.com/s/0bzikt9mnl4tr6v/STS_2.0_NoFilter.png?dl=0

Note that I've extended the range of the FFB & steering torque plots out to double NominalMaxTorque in order to show that whilst FFB gets clipped to 1.0 steering shaft torque can go above NominalMaxTorque (which was at 6.0Nm for these tests).

 

If it's done that way, the brief test I did either doesn't make sense or I did something wrong. I reduced the NominalMaxTorque by 10 (9.5 -> 0.95), and set the FFB Mult to 0.10.

Your step 2 in that case should have lowered the calculated shaft torque, which then would have been scaled up by the lower NominalMaxTorque in your step 3, returning the output to what it was with the original values. That didn't happen for me.

 

When I have the time I'll repeat your test and log some data.

Edit: Tried a very quick run with Nominal at 0.6 and Multi at 0.1.

The SteeringShaftTorque trace doesn't change, suggesting that the multiplier is indeed applied later in the chain.

The FFB output sure does change - it now maxes out at 0.435 for an input torque of 10.816Nm.

That's not immediately making much sense to me, will give it some proper thought over lunch.

 

*Bump* in case Lazza missed the above edit. I probably should have made it a new post.

 

Thanks, I had read your post and was happy to wait for your test Will await further testing... not really in a thinking mode much myself lol

 

That mechanism (if true) would perfectly explain my ffb multi observations.

So whilst NominalMaxTorque never actually changes, lowering the FFB multiplier has the same perceived/observational effect as if having raised the NominalMaxTorque value.

For example, if say NominalMaxTorque = 6.0Nm and in that car you drive you experience an absolute max virtual steering torque of 12Nm (when driving at the limit in the high speed corners). Clearly the ffb wheel is going to be clipped for half the upper virtual steering wheel torque range. Then using an FFB multipliter of 0.5 will drop the max virtual steering torque from 12Nm to 6Nm (by halving all the calculated virtual steering wheel torques from the physics engine) which is then below the NominalMaxTorque value, which would hence avoid any ffb clipping issue.

 

I've just tried a run with Nominal at 12 and FFB multi at 2.0. This time it did what I'd expect, pretty much... the steering torque & FFB traces overlay almost perfectly, it just clips as torque hits 6Nm. There's some (large) discrepancies when the vehicle is static that I don't remember seeing in earlier traces (zero speed multi is at the default 0.3).

Strange that doubling both Nominal and Multi works as expected, but dropping both by a factor of ten doesn't. I'm going to try a run dropping both Nominal and Multi by a factor of two rather than then, see what happens.

 

I also tried a run with Nominal at 3 and Multi and 0.5, it did what I'd expect (steering torque & FFB traces match, clips at 6Nm).

So then I tried Nominal at 1.5 and Multi at 0.25. Here it doesn't match, I see 6Nm only produce ~60% FFB output.

At the moment it doesn't make a lot of sense to me that 3/0.5, 6/1.0, 12/2.0 all do what I'd expect but 1.5/0.25 and 0.6/0.1 don't.

 

Paul, if you read on to Tech's post after mine you quoted and responded to, he gives a mechanism that explains my observations. Exact same result just different method to achieve it (and it's perceived effects).

 

So Lazza, would seem we were both correct in our observations, lol. Just that yours starts happening when you drop the Nominal and FFB multi beyond a certain point.

Looking forward to Ade finding out what's causing the anomaly/possible-bug you observed with Nominal = 0.95, FFB multi = 0.1.

 

Indeed, I'm happy for you that the FFB mult doesn't quite work as I thought

 

Ade found out what's causing the anomalies! It's the "steering torque capability" function which is set to 2.5Nm by default.

So if you lower the NominalMaxTorque below the steering torque capability's 2.5Nm value, the otherwise consistent pattern of how NominalMaxTorque and FFB multi interact to produce the max steering shaft torque value that corresponds with max/100% ffb output, breaks down.

Take for example the ones Ade confirmed work as expected. 3/0.5, 6/1.0, 12/2.0...they all produce a max virtual steering shaft torque of 6Nm that corresponds with 100% ffb output signal. However, go below 2.5/0.417 and a 6Nm steering shaft torque no longer correlates to 100% ffb output.

For example, Ade's 1.5/0.25 test gives an ffb output of only 60% when the virtual steering shaft torque = 6Nm. You can see why this is so by dividing the NominalMaxTorque value by the steering torque capability value = 1.5/2.5 = 0.6 = a 60% ffb output signal.

For (Lazza's) 0.6/0.1 test, when the virtual steering shaft torque was at 6Nm it would have produced a 24% ffb output signal (because 0.6/2.5 = 0.24 = 24%). etc.

Lazza, this perfectly explains why the max torque output from your ffb wheel felt severely reduced for the same virtual steering shaft torque when you tested 0.95/0.1.

Lastly, for example with Ade's 1.5/0.25 test which produces an ffb output of only 60% when the virtual steering shaft torque = 6Nm, it means 100% ffb output now occurs at a virtual steering shaft torque of 10Nm (instead of 6Nm). For Lazza's 0.6/0.1, 100% ffb output occurs at a steering wheel torque of 25Nm, which again makes it very clear why it felt so much weaker in ffb for you vs 6.0/1.0. If you had tested no lower than 2.5/0.417, it would have felt exactly the same as 6.0/1.0.