Physic discussion thread

Really good data here.

A question though: The feeling I get is that the car kind of slides/hops and then suddenly catches in a corner. Are you saying that the diff isn't opening up and this may be the cause of that feel/behavior?

I'd be curious if other sims like ACC handle the diff like they do in AMS as well.

 

No guys, don't ignore him, you'll be doomed forever with your useless life.

Back to the topic.

100Nm of viscous lock are almost equivalent to full lock (same as similar value of preload), so in some cars I'm quite confident there are leftovers not even shown in setup menu.

Renato said it was not active anymore but I can confirm it is... and doing all the best for ruining your driving.

 

Here is another test that I wanted to do since @Avoletta1977 brought it up. I noticed you mentioned the Group A one tire fire (inside tire slipping a lot under acceleration) in another thread. I definitely feel that in many cars. I'm personally more interested in lack of turn in, but I'm not a fan of one tire fire either.

To anyone not sure what one tire fire sounds or feels like, it's when you accelerate out of a corner and the revs shoot up while one of the rear tires screams. You lose quite a lot of acceleration potential, because of it and instead of oversteering into a drift, you generally end up understeering. Some might actually prefer this for stability though. All drivers are unique.

I compared AMS2's Group A Mercedes-Benz 190E to R3E's version of the same car. I even used the same silver Keke Rosberg livery for better science!

Track is Hockenheim Short Layout B.

Both sims used their default setups. I went expecting R3E to show more rear tire difference off throttle. However the results were a bit surprising.

Here's AMS2. Same old story as we see with the F309. Rear tire speeds do not show any difference when braking and coasting. Only exception is that last dip around the 2200 vertical line. But you can definitely see one tire fire. It's those peaks where the other line separates from the other while both lines are going upwards. Interestingly you do still get some amount of throttle oversteer during exits.



Here's R3E. Surprisingly similar story when you look at the dips. Very interesting. But there is some difference all the time. I wonder what kind of effect this would have in AMS2? Probably pretty nuanced. Anyway you can tell I applied the same driving style I did in the first F309 graph: stabbing the throttle around the apex. Not as much though. This keeps the car rotating and maintains the speed somewhat. The bigger difference to AMS2 is the lack of one tire fire. It looks pretty well behaved and throttle oversteer feels more dangerous. I'm pretty sure many people actually find this a crazy setup and would prefer some one tire fire if it made exits easier.



Here's another lap where I drove very smooth and "by the book" on purpose in R3E. It's nowhere near the previous time, because the car simply bogs down while waiting for it to turn around the tight corners at Hockenheim short. So maybe the car is simply meant to be driven with some throttle oversteer already at the apex. It would definitely not be the only one and the tires do facilitate that nicely being pretty forgiving in both sims.



I think at the end of the day Group A is not the reason to look into diff opening earlier in AMS2. I'm pretty sure that would only make it feel a bit more responsive and a difference like that is easily lost in the . But if it does suffer from the same snap phenomenon (where it snaps open suddenly and not gradually mid corner) then that would be something that makes it difficult to drive.

A more interesting thing to look at is that one tire fire thing which it has plenty of. You can increase the amount of clutches to increase overall locking, but I haven't tested if it helps. Still, I would not want to do that when the differential is already way too locked.

---

I'm not one to talk about this too much, but I feel like this is the problem going forward: multiple updates have reduced locking per clutch in the clutch LSD. This has introduced more one tire fire which makes sense since you have less friction. However it hasn't really solved the always-locked issue. Yes you eventually get a more open differential at very slow speeds, but most of the time it's just stuck closed until it opens with a noticeable step causing a code brown if not a spin.

 

Yes and I only now noticed you ran the same test as me in AMS2 around Hockenheim short . Thankfully we compared different sims so it's still interesting.

Yeah I figured it's setup for drivers who wear cool sunglasses. And indeed int would probably be pretty much impossible to bring the differential setups close. But a similar feel can probably be found. I'm not sure what this would prove though. Seeing even a small amount of gradual unlocking off throttle tells a story.

Here's what I was thinking as well: is a clutch LSD actually supposed to stick a lot before breaking open? Do you feel a step? It doesn't make sense to me that it sticks with very low preload and practically no ramp - or 89 degrees meaning torque doesn't turn into clutch pressure if I understand it right.

This leads me back to your question: does it matter? Are we using the wrong type of LSD anyway for many of the cars? I don't mean this from a purist point of view, but rather from an over engineering point of view. I respect the heck out of complex simulation, but it'd be kinda funny if Reiza found a fix for this and then later ended up making it so that only a few cars use the clutch LSD.

 

Ok since @CrimsonEminence asked, here's another test with the Mercedes-Benz 190E (Group A/DTM) at Hockenheim Short Layout B.

This time I used a setup that's my best guesstimate of a matching differential setup to the values in AMS2. So the values, not how it actually ends up working due to the glitch we may be seeing.

It's difficult to do this, because many sims use percentages for power and coast where as Madness clutch LSD uses ramps (as it should for a Salisbury type clutch LSD). I have no idea what type of differential is supposed to be in there, this isn't about that.

The default differential setup in AMS2 is as follows:
Preload: 20Nm
Clutches: 4
Power ramp: 74deg
Coast ramp: 58deg

I have no idea how to take clutches into account when matching the setup with a system that only uses preload (Nm), power-% and coast-% But 89deg is maximum and higher ramp value means less locking. So I just calculated that 74deg power ramp is about 18% power-% and 58deg coast ramp is about 35% coast-%. I'm already thinking that this is just a fun experiment and not applicable in any way.

Thus there were the numbers in R3E (I used the closest available which were really close):
Preload: 20Nm
Power: 18%
Coast: 35%

---

Here's R3E. The one tire fire moments under acceleration aren't exactly similar, but neither is the lap and bumps on the track. This is probably really sensitive to differences in the tire model. However it's clear that the inside tire does like to spin more than the outside tire in multiple places.

Behavior off throttle is really different compared to AMS2 with "similar" values. Mind you tire speed difference off throttle is never going to look that contrasty compared to one tire fire under acceleration where the lines can get really far apart. But you can see the other line gradually detach from the other line. And then you have two sharp V shapes in the dips meaning that the rear tires are spinning at different speeds when turning, letting the car rotate nicely.

Talking about driving style is somewhat irrelevant this being R3E, but with this setup it's actually quite easy. You can trail brake into corners and don't have to wait that long for the car to rotate. And then just exit without being as careful with the throttle. You don't want excessive inside tire slip, but at least the car won't spin if you accidentally blow on the throttle pedal.



Here's AMS2 (same session and graph from before). Same story from last Group A test. Locked most of the lap except when the inside tire starts spinning more than the outside tire under acceleration.

I repeat that this does not necessarily prove anything, because who knows how to accurately convert the differential setup values from a clutch LSD to whatever R3E uses here. But even so it leaves us wondering what those setup values do in AMS2.



---

I also tried going for a similar differential feel. The values I ended up using were quite wild.

Preload: 20Nm
Power: 5%
Coast: 99.9%

I'm not sure about the coast as it's not that locked once it opens in AMS2. But then again it seems to never open so I guess it's alright. Anyway this was just a side experiment.

The car actually behaves quite similar to AMS2. Does not like to turn off throttle which is similar to the default setup in R3E, but not completely. It was a bit less agile off throttle with no liftoff oversteer. And then the car turned better when you were on throttle. You would chuck the car into a corner while trail braking rather than turn in smoothly and then control it with the throttle. More throttle actually helped stabilise the car while keeping it in a slide. With the obvious caveat that the inside tire was on fire thanks to all the slip from one tire fire.

 

Has anyone "in the know" heard anything new from Renato and the boys on the differential issue?

 

Mm wer'e all patient. its a long term, not easy thing, but as mentioned here by others devs/r are aware and The man has a list/lists of stuff!! apart from this also

 

Tested a Porsche mod where it is possible to change at will the differential properties and I can confirm that the best compromise is the geared LSD, by far.

Sadly the mod (I think using standard Hard slicks from GT3s) has insane slip induced drag and, no matter the 400ish hp, it's almost impossible to play with it in second gear al low RPMs (it bogs down below 4000).
This has been the plague of Pcars for years and it repeats here... otherwise it would be quite fun with some lift off oversteer and nice suspension movements.

If you download it, please consider that the changes in setup menu are ineffective and the diff changes must be operated from text file (ask in private if you want).

 

At least Reiza can try different differential options with the developer tools. If the end result is that a bunch of cars are just nicer to drive then they might prioritize it in their backlog.

I think since The Man has acknowledged it, I won't post new differential tests before there's an update. But backers can ask for my help if needed. I know these tests take time and if I can help us get that update any faster then I'm here.

 

I think when the time is right, Renato will appreciate some help from you telemetry experts. He always likes things to be further along than we do before opening the curtain, though...

 

If you want to play with the mod it's here...
Downloads Autmobilista 2

In order to activate the Torsen, find the "RECORD RWD_VLSD_rear_diff amalgam_diff" line into driveline.rg file and substitute with the following:

RECORD RWD_VLSD_rear_diff amalgam_diff
pair:left wheel_RL_internal
pair:right wheel_RR_internal
diff:shaft driveshaft_GT3_POST
difflanet planet_gear_rear
mat:stiffness 1.0e6
diff:drag "0.0e-2 0.0e-2 0.0e-2"
mat:flags 0
# Torsen
diff:gear_lsd true
diff:bias_ratio "2.8 1.0"
diff:right_ratio 0.5
# final drive
dl:ratio "drive"

The line
diff:bias_ratio "2.8 1.0"
is quite obvious with the first number for drive ratio and the second for coast.

 

I would for 1 would like to see your results alongside what we know!

 

Ok I'm out .

I accidentally wrote "update" when I meant "tests". I can test once there's something new to test, but for now it seems that the issue repeats itself in multiple cars so there's no point looking for more evidence.

Thank you for your service! It's unfortunate it has taken this long, but I'm sure being vocal in beta got the gears turning. They are just a bit... sticky.

 

I would say the first as in most of the AMS2 cars it has been solved.

The Camaro ARC was a very noticeable case in the first builds, decelerating at full gas in second gear when cranking up some steering angle.

Now is fantastic, so it certainly can be cured with some massages (I think mainly on tires).