Physics discussion thread

can't remember if it was in the update V1.6.3.x or V1.6.4.x, but the change in tire physics made a big step toward grip balance.
The rear axle had significantly more grip and the front axle only slightly less.
I felt the oversteer/understeer ratio was much better (and that was with no setup changes).
Do you feel the same way, or am I just imagining it?
If I'm not imagining it, they could make another big step like that, and if they also reduce the effect of engine braking so you can lift and coast, I'd be very happy.

 

yes, the updates after 1.6 improved the gt cars. now it is more a setup thing.

 

I don't know if this is the right place to post this. If not, then I'm sorry!

I've recently been playing around with telemetry tools and have noticed that the differential in most rear wheel drive cars with a clutch-type LSD (Especially GT3, but it happens on most RWD cars with an LSD) simply isn't locking/clamping properly when getting on throttle under lateral load?

Below are a couple of screenshots of some telemetry from a short race against AI in the Aston Martin Vantage GT3, base setup (100Nm Preload, 6 Clutches, unknown ramp angles) except I lowered the TC and ABS by a few clicks. These results were consistent lap after lap.
(excuse my awful layout, I don't use i2 very often...)
Telemetry Image

Telemetry Image

The moments you see the inside rear wheel spin up in the telemetry were consistent with the rear end becoming light and floaty, leading to doing a lot of my rotation on corner exit with just the throttle and balancing the steering to compensate.

Now, I'm not a race engineer with access to real world IMSA GTD telemetry. Maybe what I think is a fault is actually entirely intentional behaviour and the ramp angles are set really steep on purpose or something. I decided to compare these results to ACC and LMU, sims that are based entirely or largely around GT3 cars albeit with different regulations, setting the diff up as similarly as possible.
They both had the speed difference between the rear wheels while on throttle and under lateral load to be no more than 10-15rpm at the most, even when trying to drive like an absolute tool and stamping on the throttle on corner exit. We're seeing at least double if not quadruple that difference in speed across the rear axle in this case, and consistently across the board from my testing.
If the wheels broke free in the other sims and I lost control, it was both wheels losing traction and spinning simultaneously. Not one wheel spinning up massively and causing a lazy spin like we can often see in AMS2.

I think the result of this weird unlocking/slipping behaviour above 45% or-so throttle is that we will lose some traction/grip from the inside wheel and end up with a shift in lateral load or balance across the rear axle while the car is loaded up through a corner, giving us this strange feeling in the rear end, and that kind of vague slidey-ness that people often complain about with AMS2.

It just feels off. These GT3 cars produce a tonne of power and torque which, when transmitted through the diff to the rear axle, should exert a HUGE amount of pressure on the clutches, preventing most if not all slipping, dependent on ramp angles and such.

Anyway, sorry for the essay. I hope these findings are maybe useful to someone, or may spur people to do their own research into the diff behaviour in this sim.

EDIT - I've figured it out! The ramp angles are backwards in all the default setups in which I can actually view them.
The power ramp should pretty much always be a shallower angle than the coast ramp in a racing scenario.
I.e. Power 45°, Coast 70°
Currently the cars are all setup the opposite way, so per this example: Power 70°, Coast 45°

Without getting into the specifics of how an LSD works (it'd take me multiple paragraphs to even begin to explain properly...), the ramp angles are essentially grooves cut into physical plates inside the diff. When the wheels begin to spin at different speeds it causes a guide pin attached to some gears to move up the ramp and applies pressure to the clutches in the diff that are responsible for essentially locking the speed of the two drive shafts together. A shallower angle (closer to 0°) allows this pin to move further up the ramp and apply MORE pressure to the clutches. A steeper angle (closer to 90°) doesn't allow the pin to move up the ramp as far, as such there is less pressure applied to the clutches and the diff allows some slip. (An extremely simplified explanation, in no way meant to be an actual guide on how diffs work...)

Having the wrong ramp angles by default is what's causing the diff to not lock properly and make the cars behave in this weird way.
It's also what's causing the cars to be too sensitive to trail braking, making them snappy on corner entry if you trail too hard because the diff is actually trying to lock harder under deceleration than under acceleration.
Though due to how torque is applied to the diff I expect the actual values don't follow that trend exactly.

 

• Locking Action (Generic Explanation):
  Steeper ramp angles (e.g., 30 degrees) result in more aggressive locking, meaning the differential engages more quickly and transfers more torque to the wheel with better traction. Shallower angles (e.g., 60 degrees) lead to less locking, allowing for more independent wheel rotation, which can be desirable in certain situations.

Differential Tuning Discussion

It's not a bug, just an issue of terminology.

But you are correct that values never or unlikely to be used in real life are possible in the game.

 

This! Please don't let this post die without further investigation!

 

So a 30° incline is steeper than a 60° incline, based on this difference in terminology?

And yes I understand that values not used in real life are possible in game. Surely if the diff is so accurately modelled, as I've been told, then using values that are opposite to what's seen in the thing you're modelling in real life wouldn't make sense.

Now I'll admit my experience comes from working on time attack and track cars in my friend's garage, not GT3s in a professional team but I've never seen a ramp set with the power ramp less aggressive than the coast ramp. 40/65 is way more common for a looser car on entry with more stability and confidence on exit.

Either way, that's just my thoughts. if you guys don't think it's a problem then that's fine, I won't hijack this thread trying to argue my corner. Cheers!

 

I'm not sure what you are saying, because smaller number = less locking in those coast / power ramps and like you wrote yourself, closer to 0 degrees would mean more locking and that's how it is in AMS2.

The cars where you can see the power ramp and coast ramp values (and often with something like power ramp 70 and coast 40) have usually even more powerful engines and if you would change the power ramp to be smaller (more locked), you would just have more potential for additional rotation and oversteer when going on throttle. F-Ultimate is a good way to try it, if you lower the power ramp value (more locked), you get more oversteer issues when going on throttle (when applying throttle while still turning).

ps. I'm still not expert in this area, but in the past I had a wrong understanding about diff and how it effects on power and off power. I thought more locked = less rotation in all situations, which is not the case. So anybody that might have that same same incorrect understanding, more locking on "coast" (so during braking / coasting) = less rotation (more stable) BUT more locking on power = more rotation (the outside tire get's more power instead the power going to insured tire, the path of least resistance)

 

Ok I understand what you're saying, but I guess my problem isn't with the actual diff behaviour, or the way it's modelled, but the way it's set up instead.

It's behaving exactly as it should.

However, with the way you have the ramps set currently we see the diff more aggressively trying to clamp shut under braking and deceleration. Sure, this makes the car more stable on the brakes but can lead to unexpectedly sharp rotation if turning while trail braking. Largely it's fine, though.
Coasting behaviour looks and feels fine tbh. The diff opens up without the torque from the engine driving it, allowing for nice and consistent rotation.
I suppose the problem comes when transitioning between coast and power ramps when getting on throttle, because the power ramp is so steep (I'm unsure if our terminology is clashing. By steep I mean a physically steep angle cut into the pressure plate, where the guide pins for the planetary gears rests.) we exceed the relatively limited clamping force afforded by the steep ramp and the inside wheel begins to slip in a way that feels like it comes on too early, the sudden 'unlocking' leading to the behaviour I noted in the previous post. A vague, washy rear end that tends towards a slow wide oversteer arc from the rear end.
Like the car begins to rotate around its own center.

In every real car I've worked on for track and time attack, I've never had a car run a coast angle that's less degrees than the power angle. (I'm trying to explain this in neutral terminology to avoid potential confusion between steep and shallow ramps.)
45/60 (pwr/coast) is very common for tarmac racing, even up to a pretty aggressive 40/80. But I've never seen 60/45.

I suppose as someone who has transitioned from real world club level Motorsport to the sim, I'm somewhat baffled by the choices made. Especially as, in my opinion, the differential is second only to tyres when it comes to importance in how a car feels to drive hard and how it handles being on the edge.

Edit - However I'm totally willing to put my hands up and say I could very well be entirely wrong. I'm not a professional, so my opinion realistically doesn't mean very much at all in this case. I'll stop taking up your time and forum space with my overly long posts (sorry!)
My opinion is out there now, that's all I intended.
Cheers!

 

I don't know anything about real cars but this is fascinating (and I will likely test it next time I can get on the sim!). Thank you for this discussion!

 

Just use a real-world balance that you think will make sense. That's what I do When looking at a set-up in the game (from whatever source), one of the first things I do is get rid of extreme imbalances in diff, dampers, ARBs, etc. Even if in the right hands those set-ups are able to scrape off some lap time, I already know it's not a car I want to drive or will find comfortable or intuitive in the long term (context of an actual race).

AMS 2 is not perfect, but most settings will yield what they should in real life, so go with what you know and prefer and you'll likely end up happy and closer to the front at the end of a race.

 

Will reiza could revise the default setup of the gt3 gen2. There are so many things that make the handling way better.

 

The engine braking is to low. It make all the difference to put it in 8-10. For the rest, i am not really good with setup. I think increasing the preload a bit make the physic better. And lower the clunch from 6 to 4 also make a difference. But, there mostly somebody more good with car setup that could explain at the dev here what could be change.

 

Reducing engine braking (increasing the number) is a must for me - but that's my fault, I tend to downshift WAY TOO SOON, a bad habit I can’t seem to get rid of.
Reducing clutch plates increases stability, but makes the car hard to turn (as expected). Then, lap times go down the drain.
For sure, changing car setup does have a huge effect (as it should) on cars perfomance/behaviour and we should take the time to understand it. But many will judge the sim using default setups results.

Although I’m an engineer (electronics and computers), I don’t have enough expertise to discuss mechanics at this level of detail. What I can say is that I have some experience throwing around 1–2ton objects with four wheels attached, and I think the "problem" with AMS2 isn’t the complexity of the simulation, but the perceived result. It’s just the small details that others seem to do slightly better.
For example, when driving another sim I hadn’t touched for a long time, I immediately recognized physical behaviors that better reflect what I feel in a real car. The rear tracks the front axle much better, and when it finally steps out, the car loses a lot of momentum or acceleration (I think AMS2 handles the latter well enough, but not the former).
Some changes in FFB scrub effects made it very obvious that the rear is always sliding, even on high-downforce cars, and that we must always control the car’s rotation using the sliding rear rather than a grippy front (hope this makes sense).

I also tend to notice the lack of inertia consistency — sometimes there’s too much (leading to the "floating effect", a "never ending" slide), and sometimes too little (like with bumps and crashes, where cars often seem like very low-mass objects with massive accelerations not justified by the forces involved).
AI cars suffer from the same problem, leading to cognitive dissonances (which translates into the sensation of “this is not real” or “bad physics”).

 

Is increasing preload help for gt3 gen2

 

Increasing preload, reducing clutch plates, etc., will make the car more stable and less prone to rear sliding, but a racing car is all about making it as unstable as possible while still retaining its driveability (which depends largely on the driver).
The Red Bull F1 Racing Team can speak about this subject with more authority .

But on engine break subject, the truth is I also change the brake bias by 2 points rearward… which is the same as saying I’m not relying on engine braking as much, but I can use a little more rear braking power to compensate. Without adjusting engine braking, I often spin due to excessive rear-end braking demand. That serves my driving style better and it's realistic.

 

Even though I agree with setups are very subjective and you can't please everybody or make them work ideally on every track, I do think the default setups with current AMS2 physics are quite often very far out of the norm what you find in other sims, more oversteery than understeery and make trail braking deeper into the corner often quite tricky.

IMO AMS2 would benefit from having safer default setups for popular classes especially (even at the cost of potential pace). It would make the cars and physics feel less slidy/floaty/drifty and cause less furstrations with players new to the sim if the default setups would have rear that goes so light so easily in corners. The cars ofter rotate still well enough even if you shift the balance more towards the rear.