Formula-Retro drivability

why is there no one online?
I will be testing tonight anyhow, I noticed a difference in TT yesterday quick test

 

edited

 

Interlagos H long
Beta test felt better off the bat to me, then followed by main same settings and in both the Brabham is a different beast and much improved since diff and camber changes in last update.
Default, setting in race and just did TT in main went from 2.35.7-- to 2.32.41
Hard to pin down a big difference but beta times were faster by .300, (may have just been my driving)

Seems much improved to me, have you tried to beat your old TT times?
I would expect all the retro times to drop by several secs now.

 

Sounds good, like I said I just used default as reference, asI know it well in the Brabham and it was a tricky but drivable car. now its tip top and a world of difference imo. If it was like that 2 months ago a lot of threads would have been shorter. Castor tweek caught my eye in last update aside from diff.
Thanks Reiza for making it more of a pleasure. Imo.

 

Guys I spent some time today messing with it, and at least in the F-Retro Lotus 72E, I can feel NO DIFFERENCE between:

Preload: 0
Clutches: 2
Coast Ramp: 89
Power Ramp: 89

and

Preload: 0
Clutches: 2
Coast Ramp: 25
Power Ramp: 25

Even with preload at zero, and coast ramps as close to 90 as the game will allow (PC2 allows 90, so this is odd to me), I could still do two-tire burnouts.

Either setting made ZERO difference to the handling of the car. For my last back and forth that I THOUGHT I had video captured, (I'll try again tomorrow), I did 2 laps with each setting back to back and got the exact same laptime on my second lap both times.

Guys I dont think these settings are doing ANYTHING at all.

They DEFINITELY did with the geared diff. I had to constantly tighten up the coast side of the geared diff at Imola 72 to keep it from immediately spinning as soon as I got off throttle into Castellaccio (Acque Minerali). You could feel a distinct difference between 1 and 2, and 2 and 4 (4 was the middle).

The clutches dont seem to affect anything.

And the understeer is still massive.

With that said, I love the way the car feels, and the tires feel. It's just the insane understeer that is impossible to live with. They just DO NOT TURN. Corner exit and other high speed corners are just plain fun and feel good. But the understeer at low speed due to the diff is unbearable.

ALSO!
I did the same thing with the F-Classic G3M3 and the FV10G1M1, and I could not feel any discernible difference in car behavior with those either, even with ramp angles at absolutely extreme settings.

Something is plain broken here. I would have just assumed I cant tell what changes diff settings make ,but we all experienced how radically different the car was with the geared diff, and the geared diff responded very well to adjustments.

 

If people are working on it then I'll leave it alone.

But I will say I went back and forth on the 49C using max preload (which is 600Nm!), max clutches, and minimum ramp angles, and using zero preload, 2 clutches, and max ramp angles, and I honestly couldnt tell any difference in how the car behavedl, I would just do two laps one way, then two laps the other way back and forth. I set a bunch of soon to be beaten world records at Imola 72, that way, though! Neither setup seemed quicker at all. I just got better at driving the car, and the car behaved the same.

I also have to say, man these cars are handling great right now! The F-Retros have their insane understeer that may or may not be due to the diff issue, but other than and the getting spooked on crests, they are feeling great too. Great fun to slide them around, just like the F-Vintage.

 

I have no time in this to speak of so cant comment, but it does not supprise me that all things are not correct across the board.
The actual settings and what they do/dont do/supposed to do, have been confusing since release of these 2 classes, its a shame that things are still in a kinda limbo esp when simple wheel on grass test shows clear problems, (I have not repeated since first test last week).
I can only comment on the Brab as having more grip both ends, less over/understeer now.
Prior to last build I ran the geared and was like you chaps thinking it was so much better.
Brabham now feels to me ready for some wet driving/storm racing!, not possible before last week.
I will have to try the others.

Agreed Sooo much more a pleasure than a pain, can only improve with

We know that changes across the title have been ongoing and using onboard feedback from all areas.

What struck me yesterday was the ease at which I broke mine, even during a race, and yep I tried on old build for a lot more laps, quite fun really and more time can be shaved im sure.
Fancy running some laps (race) tonight!

 

I went back and looked at my video captures...

I was primarily running at Imola 72... Let me know if you guys can replicate or concur. With preload at max, clutches at max, and ramp angles (power and coast) at minimum, the Lotus 72E understeers through Tosa at Imola 72 until you apply a little power oversteer.

While the rest of the corners felt identical no matter what setting, I think I found in Tosa where you can tell somewhat of a difference. If you then set preload to zero, clutches to 2, and ramp angles to 89 (This should be a VERY loose differential), you can then turn through Tosa, and although the car still aggressively understeers, about halfway through the corner the front does seem to finally grab just enough to reach the breakaway torque of the diff, but then it goes right back to understeering.

I dont think the problem is that the clutched diff is accidentally set to "spool" or anything. I think there's some fundamental issue where the baseline, loosest setting is so tight and so stiff, that it's functionally the same as a spool in most scenarios.

So if you have a loose diff, it will then be very prone to sending all power to the spinning wheel, because thats the path of least resistance. The clutches and preload pressure and ramps have to overcome that to lock both axles together. If there is no locking action, the tire with no traction will spin. If the diff is unlocking at all (which is key to tuneable clutch diff behavior, being able to induce independence between the wheels in whichever scenarios you desire), then you should see one tire spin if there is a traction disparity large enough.

To induce the largest traction disparity, I took the 49C and put all diff settings to as loose as possible, and parked it one side on pavement, and one side on dirt at Ibarra. With the loosest settings allowed in the game, the power should get sent to the wheel in the dirt, and the car should not move very much as the wheel with traction shouldnt spin much, if any.

Differentail Unlocking Test:
Assumption 1: With a loose diff, or even a medium tight clutch diff, normal driving (that is, taking a corner) should provide enough breakaway torque to allow the tires to turn independently.

Assumption 2: If traction disparity from one side to the other is great enough, we can force the wheels to unlock, as a loose diff should allow the tire with no traction to spin.

Procedure:

1. To induce the greatest traction disparity, we can put one rear tire on pavement, and one on dirt.
2. To encourage unlocking of the diff, we will set the diff as loose as possible.
3. We will then rev to redline and dump the clutch to observe whether power is sent to both wheels despite this traction imbalance.

Pass/Fail Criteria:

Pass

• Tire with no traction spins, very little torque is sent to tire with traction. We can observe a lot more wheelspin on dirt side.

Fail

• Both tires spin, leaving burnout marks, car should pivot away from higher traction tire, as traction levels are modelled correctly, proving one side has no traction and the other has high traction. This proves the diff is not unlocking even in ideal conditions that encourage it.

Inconclusive

• Traction is equal on both sides, car accelerates straight. This would indicate a modelling problem with traction levels from pavement to dirt or across each side of the diff, and would be an inconclusive result, but would indicate the problem may not be the diff modelling.

Findings

Both tires spin aggressively, car veers away from the tire with traction.

Conclusion
-We know the game is modelling pavement vs dirt traction correctly intuitively ,but the result being that the car pivots away confirms this.
-Even in ideal situations to encourage it, the diff is not unlocking perceptibly
-This would induce spool-like behavior in all scenarios, such as understeer on corner entry, tendency to oversteer on corner exit, slightly unpredictable behavior in transient scenarios, predictable behavior when both tires lose traction at the same time, unpredictable behavior over crests when traction levels decrease and increase rapidly.

I think this is happening in every car in the game, and I think this explains a lot of things that we were all noticing with other cars, and it explains why the geared diffs felt so much better, despite the main advantage of clutch diffs being tuneability. I think this leads to the cars feeling "nervous" and "disconnected" as the rear tires are always fighting each other for traction in every car that uses this type of diff, leading to odd behavior.

It also makes sense that this would be disproportionately obvious in the F-Retros due to the combination of high grip, light weight, front to rear tire size discrepancy, and short wheelbase, and would lead to unpredictable losses of rear traction around sweeping high speed corners necessitating a seemingly unbalanced amount of rear wing to maintain it, and cause severe understeer at lower speeds when there is no front downforce to compensate for the disparity in front and rear contact patches.

Again, I know Reiza is working on it, and I am sure they'll find a workaround for us. But I just wanted to present my methodology, my reasoning, and my results for review and research purposes. I actually think this gives us reason to be very optimistic for the future of the sim, as this would remove a lot of odd behaviors that we have noticed and should lead to a drastic improvement across the board once resolved, if my speculation that this is a universal clutch diff issue is confirmed.

 

Your testing method and assumptions sounds reasonable.
Lets just hope these diff issue(s) is not something that is hard coded in the pCars2/Madness Engine.

ByTheWay: Oh my old brain did just recall that I have commented on this issue before
Does same engine = same physic?

 

I think my experience with how nicely the cars behaved with the geared diff proves there is an easy workaround if it is truly unsolvable. I know some get hung up on the "simulation" aspect, but we dont need to simulate flame fronts and combustion dynamics and torsional vibrations of various engine configurations to realistically model engine output. We can approximate the nature of engine power without needing to model each individual facet of how engines work. I think we can use the geared diff as a placeholder for how limited slip differentials work. There will be slight differences to how a clutched differential works in reality, but in game, the clutch differential is, to at least some extent, broken, and the geared diff is going to be much closer to the reality of how a clutch diff operates than the current clutch diff modelling.

I get the impression that given the power of the madness engine and the SETA model, Reiza wants to be that company that simulates as many facets as possible as correctly as possible in reference to a common model. I think they may very well pull it off in a way that SMS could not. But I think in the event there is no workaround in the near to mid future, I think the only responsible thing is to allow geared diffs to be used in all clutch diff cars.

They will figure out something.

 

Summary of the wall of text:

Differential implementation is broken for some reasons (we already knew it but I was too lazy to use bad softwares to prove it).

Devs have much better tools for understanding why and solve it.

Hopefully they will keep us in the loop while doing it.

End of the story.

 

Your assumptions are correct BUT it also depends on how the engine inertia is modeled into the driveline equations.

I've seen plenty of mistakes in this specific aspect...

Let's wait and see what Reiza will find.
If they need any support we are here...

 

I think the pessimist in me always wins out, but honestly I do agree with you. I think this should be properly solvable, and I appreciate your optimistic outlook! it adds some perspective!

 

It is not rocket science. If they have access to the source code of the physics engine (as I understood) this should be an easy fix.

 

I think this is cause for celebration, to be honest. This means that one way or another there will be huge improvements in the pipeline for fundamental things. This is low hanging fruit. They'll fix it.