Physic discussion thread

Respect to baseline setup there are some other modifications in the video:
- diff open (it was just for testing... you can experiment at will but avoid geared diff as it's disabled even if present in the menu)
- a bit softer front and stiffer rear both for anti roll and springs.

The modder @gvse has done an impressive work on that car and the only issues are graphic glitches on tires (too flat and colliding with wheel arches).
I have done nothing at all.

It moves beautifully and with some small updates on tires and diff from Reiza it could easily be the best vehicle in the game (and demonstrates why modding should always be supported...).

 

TBH while these observations may be very pertinent with the experiment of the doughnuts to me they also feel the most irrelevant for the normal driving diff issues. Without being polemic that is.
For example right now if you drive through a relatively tight corner (say the hairpin at Suzuka with a GT1) with medium to low preloads and standard ramps the diff remains locked pretty much across the whole corner loading you with a sh..load of understeer until you exceed a certain level of steering input when it snaps open and your car darts towards the inner wall. This all happens with little to no applied longitudinal load as you are coasting. We are nowhere close to "longitudinal grip after peak". Actually the problem seems to be worse if you go through the corner slower than the car lateral limit would allow (hence with theoretically higher longitudinal available grip).
In many cases, if you expect it in advance it is enough to give a jab to the throttle to shut the diff again and regain control of the car which is another oddity: any smallest throttle inputs seems to immediately trigger diff closure and tons of understeer even the shortest jab.
None of these points would have any significant effect in addressing the issues we are having in a condition within the limit of tires slippage or thereabout. IMHO I would very much prefer to have a properly working diff around the max grip limits before one that works correctly while far into macro-slipping conditions since I usually go around tracks and don't spend time with doughnuts or Ken Blocks drifting stunts.

 

I have noticed this behavior with the Mercedes GT1 car, around low speed corners, the car will suddenly go from understeer, to turning in wildly and usually spinning around. Have you figured out what diff settings to change to counteract this? I tried setting the coast setting down about 10 or 20 degrees but it didn't help. Is this a preload setting that has to be changed?

 

I don't think it is about driving like Ken Block or doing doughnuts (though that is obviously effected) but when pushing cars to their limit and really trying to achieve maximum times on the leaderboard (for example) you do tend to have the car sliding on the edge of the tyre limit and this "problem" is really evident and off putting, as is the diff not working as it should.
I believe this "mod" to the Porsche mod is just a proof of concept really to show where the problem (at least part of it) can be found.

Maybe...I'm just thinking out loud here

 

Of course!
I don't care if a car can't do donuts...
But if it can't, it can't properly power oversteer at the limit in second gear as well.

Now we are forced to use very tight diff settings for making the car rotate and this, by consequence, generates pretty obvious side effects.

I just wanted to demonstrate that using 'softer' methods (inertia) for avoiding bleed out of power due to inner wheel crazy spin, the car becomes much much better even with an open diff.

So we got two concurrent issues.
Wrong tire curves (on some slicks) that are made believable only by an also quite broken diff...

Below those issues there is a great sim.
Let's hope it will emerge.

 

Yes I do. Solution is playing with pre-load till it disappears (or actually you keep your diff locked at all times drifting laterally your rear a lot).

 

I agree the solution is not far. Probably tuning some stuff around will resolve it. Still while tires curves after peak may be less than optimal as you are saying, most our even record setting laptimes driving does not involve high longitudinal rear tire loads when the diff is particularly misbehaving which is just after turn in and at the onset of corner exit so together with tire grip curves something else in the driveline must be corrected IMHO and it is apparently not in the diff calculations per se since it appears to open up at some point, just not in the right conditions and at the right time.

 

I like the feeling of the tires at their limit. But I do agree that there may be too much sliding right now. I wonder if this is compounded by the differential fighting back and causing roughness in the rear of the car? The tires aren't allowed to spin at their correct speeds so the rear tires are never truly settled while cornering.

Project Cars 2 tires are often noticeably more prone to sliding and yet the geared LSD made the McLaren F1 LT feel more on rails with it when I tried it out. Maybe not at the limit since the tires are slippery, but you don't get the same slidey impression just by driving through a corner fast. The unreliable clutch LSD implementation causes similar nervousness in corners where the car is always changing its mind about where it wants to go: first it oversteers and then it understeers.

I suspect it's the other way around in terms of cause and effect. Obviously you beta peeps might now more than me about how it is. And I know only the basics about tires (Heusinkveld videos etc.). But here goes.

If these otherwise super grippy tires were "tighter" and had more grip after their longitudinal slip peak, the cars would turn so horribly. Forgive me if I'm thinking of the wrong aspect here or misunderstanding tire parameters, but I mean the way the tires allow quite harmless sliding forwards and backwards at the moment.

My suspicion is that many grippy cars rely on the tires having some leeway. This is why it's interesting to see if a functional differential will change tire design. I expect that Reiza has been pragmatic: you just work with what you have and what you know until you learn something new. This new engine being full of unfamiliar concepts, but fortunately less and less each month.

Although why this longitudinally loose tire design pattern would then extend to open differential cars as well, I don't know. Consistency? Anyway, my point is that maybe this leeway works both ways and introduces more one tire fire. It's easier for the inside tire to start a burnout.

 

To be more precise, the 79 opala is fine, but the oldstock and 86 are more lazy. I can still doughnut them at the exit from the pits on the montreal circuit (of course going over the red bit on the tarmac). Sorry for the mixup with the 86.

 

@Avoletta1977 I tried the tire mod and felt pretty nice with the clutch LSD. I also tried switching to geared LSD. Not sure it worked, cause I just got all the one tire fire in the world. I added it to the driveline file (you instructed around page 6 or so) and disabled clutch LSD in setup. Am I supposed to also enable geared LSD in setup? I used power 4:1 and coast 1.6:1 in the driveline file.

I think I'll just wait for Reiza's move and continue once we have some meaningful changes.

 

Am I the only one here who thinks rain/wet physics are exagarated. It turns into just "floatista".

 

Yes and it happens pretty quick.

Hmm I would say you get understeer rather than the expected throttle oversteer. This is what happens in open differential cars as well in AMS2. So yeah ends up feeling like the front end washing out. Depends a bit on the car and its tires I guess. Which car are you thinking of?

As Avoletta said the amount of understeer vs oversteer with an open differential in AMS2 is probably a bit wrong due to tire curves or some other factor. Maybe there should be a bit more throttle oversteer at slower speeds even if the inside tire is on fire. Probably not a universal thing though, but rather car dependent.

Yes.

I have focused more on the sudden opening when coasting / off throttle at lower speeds and unbreakable full lock at high speeds. But the non-gradual feeling probably applies to power as well so it transitions from full lock to open without much of a warning when accelerating hard. As it is it's much easier to open the differential with engine torque than it is with cornering forces (tires needing to spin at different speeds when turning).

The big picture I have about this clutch LSD issue in my head is this: there is a weirdly large torque range that cannot break the differential open from it's fully locked state. But when it does open, it seems very open. It's impossible to achieve a balanced differential setup with this in mind, because power could use a lot more locking and coast a lot less than currently. I might be making an assumption and this only affects off throttle behaviour. The amount of one tire fire could be some other factor(s).

Something in the clutch LSD or driveline logic or tires makes it so that it's very hard to break static friction between clutches and gears (simplified structure) so the gears can't rotate. They need to rotate for there to be different tire speeds so the clutches must slip. Without slip the diff stays locked when cornering. We don't know if that's what actually happens looking at it from the outside. Who knows if static friction is even a relevant factor / being simulated. It's just one way to picture the situation.

Amount of resistance in the clutch-based limited slip differential mechanism is a combination of the number of clutch plates, preload force, and ramp values combined with forces from the engine and tires. As Reiza incrementally reduced amount of locking per clutch in multiple updates, it didn't end up helping balance. It just made the differential's fundamental amount of friction lower when the clutches are slipping and allowing higher rear tire speed differences when the differential does open. But it doesn't seem noticeably easier to get the mechanism to slip in the first place.

So it is just my guesstimate that this hard to break full lock is why it's not gradual. Also no chance to break it open when coasting or slightly off power above low speeds (~80km/h+). There just isn't enough torque translated from the tires to the differential mechanism for whatever reason. This leads to unmanageable understeer at high speeds. So for example you can't lift to quickly rotate the car into a high speed kink. High aero and certain tires mask this issue, but it's there.

A few possible reasons just for the sake of conversation (Reiza already knows best):
- Tires slip a few percent with little resistance when the diff is locked -> torque from cornering is low and doesn't overcome static friction
- Clutch LSD unit's logic is flawed
- Some mistake in most driveline configurations

Btw. since at least I was mega confused by all this before diving deeper, here's an article for those wondering how a clutch LSD works: Into The Red - Limited Slip Differential

Also here's a nice video:

Might as well do this, because many clutch LSD setups are too loose and cause oversteer when it opens while turning off throttle. You usually want some resistance to stabilise the rear. Otherwise it's easy to over rotate and spin. The current low speed handling issue is compounded by the fact that the diff likes to open suddenly when you're fighting high amounts of understeer caused by the differential's fully locked state. So you go from high understeer to oversteer in a blink.

Open differential cars such as Formula Vee are quite unstable off throttle. But you can learn them, because you feel oversteer increasing the more you lift. Imagine if they were as sudden to open mid corner while you were fighting understeer by trail braking and steering hard.

Needs to be said again and again: thanks to active backers who have kept mentioning this in the closed beta section .

 

I would add that the use of the locking and the benefits or drawbacks for more (or totally) unlocked ones are different in the case of heavy-low downforce-high COG - small tires cars vs Light-high downforce-low COG-Big fat ass tires cars.
While for the first the concern for the internal tire losing grip due to reduction of vertical load is the predominant reason for the use of a LSD differential for the latter the internal tire breaking grip longitudinally too earlier than the outside is not the main concern at all. The locking is rather used to contain the natural oversteering effect that the unlocked diff would have on those.
In fact unless the internal tire starts spinning due to lack of vertical load which is practically impossible to obtain with formulas, GT1s, prototypes etc. the unlocked diff will always send more torque to the tire that is rotating faster (which is usually the external one in a corner). So the ratio behind the use of LSD differs a lot in the case of a Group A car vs a Formula to make an example.

 

This is a very good point and important, because I can put too much blame on the LSD. The physics of a completely unlocked differential on throttle are easily lost on me. It is telling that AMS1 doesn't allow nearly as much inside tire slip with its open differentials. Just like Avoletta has demonstrated. If that works correctly in AMS1 then I can see why power ramp is often higher (which leads to less locking on power) than coast ramp in AMS2 setups. Natural oversteer off throttle / coasting is something I would like to control when the diff starts working correctly. Right now it's pointless outside very low speed corners since it's going to be fully locked anyway.

So going by your logic, is it fair to say that the street Camaro needs more power lock than coast lock? And then an F3 needs a lot more coast lock (unless you prefer a nervous setup) than power lock?

 

I would say it is more likely for a heavy high COG car like the street Camaro to incur earlier in issues with the internal tire than excessive oversteer from an unlocked diff. So you probably would need a pretty tight on power setup.
With a 600 kg very low COG very wide tires and heavily downforced F1 car I would say you probably keep it as less locked as you can without incurring in too much power oversteer

 

Update is out with driveline changes.

"Opened up differential adjustments for Caterhams with LSD"

Shall we take a look at some Caterham 620R graphs later?

 

Update 1.1.0.4 just came out, "reduced differential viscous hump & clutch lock ratios" is this what we have been waiting for? A smoother transition between open and closed diff?

 

It may be part of the overall solution. In beta, it did not resolve the issues. However, most cars are noticeably improved in several aspects, even if the diff locking "bug" is still not exorcised.