Automobilista 2 V1.5 Physics Development Update

I am now tackling gt3s and I have some doubts.
Out of the box I have the feeling as if they drive with a spool on the rear axle. They strongly rotate on throttle and once the corner radius is imposed, one could almost complete the turn with the steering wheel centered and acting as a balancer to the car yaw.
Such behavior resembles me a lot like super v8s in real life, which actually have no rear diff.

Second doubt: the TC gets engaged a lot, even at level 1 it triggers at each gear change and at every turn no matter what gear you are in. With a good amount of downforce and alick tires, TC should not engaged in 4th gear at 180 kph (barcelona turn 3).

So here's my doubt (please educate me) : isn't it maybe that for any reasons the tires slip rate is mis-calculated or exxagerated? This would explain the TC which is activated very often and the fact that the differential would pretty much always be 100% locked.

I tried to lower the lsd clitch number to the minimum (2) and icreasing the preload to 200 or 250N in order to grant always a certain amount of lock, but trying to force the sim not to reach full lock so easily.
Well, I had the feeling to drive a gt3 car matching very closely what I expect and what other platforms offer.
I also made a comparison video (top default setup, bottom with 2 lsd clutches and high preload. Both TC 1)
Look at the steering wheel behavior and at how faster the car reacts to direction changes (check at turn 3-4).
If I didn't mess up final chicane entrance, I would also hava had a better time.

. .. Maybe it's only driving preference, I don't know.
In any case very very happy to have found also in this case a setup that fulfills my expectations!

 

Tried it and I did notice a difference and to me it did reduce that "wants to drift a bit" sensation and improved roation during corners without making it unstable (and didn't affect negatively on power either).

The diff in general feels quite locked to me on the default setups / values and I often end up trying to lower it, but it's definitely not fully locked with 6 clutches (just increase the clutches to max and try... there is a major difference).

 

A porsche is basically a FWD car with 6 reverse gears isn't it? That you drive backwards. It's a silly idea that they should have abandoned decades ago but stubbornly stuck to. Now computers have made them drivable ze Germans act like they're great engineers but they could have just put the engine back in the front in 1972. I'd be more surprised if the TC didn't come on.

I thought the GT3 were fine. The setups seem flawed. Too much brake power for a cheap pedal set so I drop that to 80-85%. The one I tried, I think it was the Mercedes wouldn't turn into the apex. 2 clicks of front downforce fixed it. I've noted others suggesting dropping clutches.

Everything I've driven that doesn't have a lot of setup options seems to drive really well - but I guess a lot of these are easier to drive anyway because they have less power - but that suggests that it's just down to tweaking the setups.

 

Got a weird situation on track at Hockenheim. Real weather, bit wet on track and sunny. After 8 laps it looked to me that the puddles increased in the last corner before start and finish straight just before I arrived that place.
I know there have been some issues with the championship mode in the past. Anyone else faced this too?

 

Yes, I think I've seen this happen in my championship races too. Could be a visual issue, as the track dries out and the puddles become more visible?

 

If I may, I believe the LSD works a little differently than depicted in this graph.

A preloaded LSD with a certain preload torque (say 100Nm) doesn't allow any wheelspeed difference between the driven wheels, until something forces the two rear wheels to rotate at a different speed, with a torque greater than the 100Nm preload torque. What causes a difference in the torque between the two driven wheels is usually a combination of a speed difference between the inside and outside wheel in a turn, and also grip differences between the left and right tyre.

After the preload torque is overcome, the wheels can have an actual wheelspeed difference, where the LSD tries to reduce this wheel speed difference. This braking torque between the two wheels is then the preload torque plus whatever the clutches add, which is a torque that increases with wheel-speed difference.

If you wanted to put this in a graph, it would be brake-torque of the diff between the two driven wheels on the vertical axis, as a function of wheel-speed difference on the horizontal axis.

 

The TC HUD/dash indicator in AMS 2 is not optimized. It flickers the same whether there is 1% or 99% TC activity. So, at present, it acts more like a "TC is on/present" light than an actual readout of what it is doing. There is a back door way to figure that out, though.

Here are a couple of experiments (assuming Authentic aids are being used and not the simcade global assist) for those who are curious:

1) Take the 911 GT3 (since that's the car being discussed) with default set-up and drive on a nice big piece of grass somewhere. Do a few doughnuts. First, notice that you can still do doughnuts with TC on! While doing them, you will hear the TC interfering with the engine output. Look at the pedal input indicator on your/a HUD. You will see it jumping up and down while you are holding the pedal steady or to the floor. That is the only indication of actual TC activity that I know of in the game.

If you are driving on dry pavement, you should rarely be able to notice any activity significant enough to be seen on this HUD display. On wet (with slick tires), it should be very obvious. Which makes sense because TC in these cars is primarily used as a "safety net" for wet driving (allowing you to get back to the pits on a very wet track with slicks where you otherwise might spin off, or, allowing you to carry on with light dampness hoping the track will dry at much less risk than without it). Any experienced driver should not need TC in GT3 car on a normal dry track.

2) Don't believe that last statement? Turn the TC down to 0 (using a control or if you don't have one mapped, in the garage set-up screen). Drive the car on the same track where the TC light was going crazy and making it seem like the car would be undriveable if it didn't have TC. What happens? The car is fine and you realize you don't need TC at all. If you are highly skilled, you'll go even faster without it. Normal people will have about the same lap time either way.

Lastly, there was some joking about the rear-engine Porsche. It can be trickier to drive if you have no experience with rear-engined/tail-heavy vehicles. But "real" race cars wouldn't put all the weight of the engine over the front axle (needed for steering!) any more than they would put it behind the rear one. Repeat #2 above using the McLaren GT3 car--mid-engined and closest to an optimal weight-balanced design (and why most purpose-built race cars are mid-engined). You should discover that the 720S relies on TC even less than the 911. It drives like a dream with or without TC. One of the best cars in any sim in the history of consumer racing sims (IMHO).

Hopefully Reiza can adapt the TC indicator display in some clever way to better represent what is really happening. In the meantime, do not mistake the over-active orange light as an indication that a car needs or is relying on its TC to get around the track.

 

There is something still that is not clear to me.
First, the preload is in Nm actually (I just noticed) and not N. How can a clutch be preloaded with a torque? Or is it maybe the locking torque resulting by the actual clutches preload?
Secondly, all what you say makes sense in static case, in other words if we have a certain locking force independent from engine torque.
But actually we have the pin pushing through the lsd ramps which eventually has the effect of increasing /decreasing the locking torque.

So, in the end, if we assume to always overcome the locking threshold to some extent, we will have a certain percentage of locking (min torque given to more grippy wheel) which at minimum is given by the preload, and then increases due to the pin vs ramps effect.
In this scenario, that graph makes sense actually.
In your scenario, if I have high preload and no ramps, then I should be "drifting" At low torque, and then stopping to drift as soon as the torque increases, but it's not what I feel.

Maybe I am getting everything wrong, I am electrical engineer, wrong branch

 

The preload torque is indeed the effective 'locking torque' of the clutches resulting from some preload mechanism. When the wheel-speed difference is 0, the clutches can transfer at most the preload-torque between the two driven wheels.
As I understand it, for an LSD with adjustable power/coast ramps, an input torque from the prop-shaft (so gearbox output, technically not the same as engine torque), adds more preloading on the clutches via the pins pushing on the ramps, exerting an extra locking force (N) on the clutches, resulting in a higher locking torque (Nm).

Now the difference between what you call static (no wheelspeed difference) and dynamic (wheels spinning at a different speed) is that a 'locked' clutch can transfer up to the locking torque, it can also be less! then there is just no clutch slip. The locking torque is the threshold where the clutches will start slipping and a wheelspeed difference can occur. This happens for example when one drive wheel has much more grip than the other, so one wheel can support much more torque without slipping than the other one. If this results in a large enough torque difference on the two wheels, the clutches of the LSD can start to slip, allowing a wheelspeed difference.

Now that there is a wheelspeed difference ('dynamic'), the locking effect of the diff is to transfer a torque from the faster-spinning/lower-grip wheel to the slower spinning wheel, such that the slower wheel receives more torque. The magnitude of this torque is determined by the properties of the clutches (friction behavior, usually increasing friction with slip speed) and the force pressing the clutches together. The diff doesn't enforce a certain locking % (how would you define locking as a percentage? a ratio of what?), it just tranfers torque between the two wheels and the speed of each wheel is whatever results from the sum of all torques acting on that wheel.

 

I wasn't joking. But thanks for patiently explaining to us that mid engined cars exist. Who else knew eh?

 

Yes I agree 100%. and thank you for explaining with clearer words the concept.
I just want to clarify that what I ment with "static" or "dynamic" was not referred to the wheelspeed difference, but referred to the fact that the locking torque given by the LSD is actually function of the shaft torque coming from the gearbox + preload-given torque.
In other words, in every specific moment, given a certain torque that the transmission is putting out, then the amount of allowed wheelspeed difference is fixed.

Let me try to put there some examples of how I see things and how I feel the porsche gt3 r stock LSD setup delivers.

Stock setup (my feeling is too much power oversteer, driving experience similar to super v8s or kart-like). At low "engine" torque (torque on shaft out of transmission), the two driving wheels are not allowed to turn at different wheelspeed. Then there is a transition in which the wheels are slightly free to have different wheelspeed (engine torque exceeds the locking torque given by the lsd), then again the locking torque exceeds the engine torque and the lsd becomes locked again (oversteery on power - high power).


lowering clutch number to 2 (no torque transferred to the loaded tire in a turn, car very planted but "slow"). After a certain threshold of "engine" torque, the wheels are free to spin at different speeds, and the torque ensured to the leaded wheel is quite low, leaving the not loaded wheel to spin and slip (e.g. out of slow corners).


Again 2 clutches but increasing preload (more torque transferred to the loaded wheel when high torque is given by the transmission, but with lot less power oversteer).
This is my preferred situation, in which there is a locking effect at medium-low engine torque, enabling the car to "push" and ensure traction, but then there is no locking effect at high "engine" torque, allowing the wheels to rotate at different speeds and allowing stability in corners when there is high torque applied.


THis is how it works in my mind, and it kind of makes sense and reflects what I feel.
I think it matches also your explanation, if I understood it correctly.

Anyhow thanks for the very informative discussion. I do think this is really a key-aspect of driveability.

 

Preload is actually important during braking, because brake torque and negative torque from the driveshaft (engine brake) sort of cancel each other out leaving little torque at the diff which would force pins against the coast ramp. This is the transitional phase you’ll also hit when coming off throttle slowly.

% of input torque. It doesn't work within the transitional phase with constant locking torque from preload of course, but in either coast or power phase it is a good way of quantifying relative locking torque. It's a crucial part of a clutch LSD's function that it increases locking torque with engine brake in lower gears for example. Hence relative locking torque in % is useful.



And here's an interesting one related to the graphs flying around here



AMS2 just lets you set preload locking torque as Nm rather than setting the preload spring rate which then affects the minimum locking torque through the friction faces i.e. clutch pack. The graph above seems to just test locking torque with two different amounts of friction faces i.e. number of clutches so same preload spring leads to more minimum locking torque.

Also take note of how locking torque only starts to increase at its max rate some distance away from 0Nm input torque and the transitional phase. Unfortunately data around 0Nm input torque is sparse on that graph, but you get the idea.

 

One thing I would add is that the input shaft torque is not in fact what forces the diff-clutches to slip and create the wheelspeed difference. Imagine driving in a straight line without wheelslip, now the input shaft torque can be super high but there is no reason for the drive wheels to spin at a different speed; there is no difference in the torque being transferred to the two drive wheels.
Only once there is a difference in torque transferred to the two drive wheels, which usually happens when one wheel has (much) more grip than the other, then a wheelspeed difference can occur if this torque difference is bigger than the (current) locking torque of the clutches (which also depends on input-shaft torque).

Effectively, there are more variables and dynamics involved than can be represented in a single graph. Also the transient between locked diff clutches and slipping diff clutches is one that really needs a wider overview of the motion of each individual shaft of the drivetrain, where each shaft will have its own equation of motion with all the torques acting on that shaft (a bit like Ohms law but for mechanics ).

 

Right, in this way a locking effect as a percentage makes sense, as the ratio of two torques. The only thing is that 100% is not the maximum per see, if you were to add even more friction plates or even lower ramp angles, you could have a higher than 100% locking effect from the input shaft. (although basically you would effectively be running in spool mode as there is no (normal) way to have the two wheels support a torque difference much greater than the input shaft torque.)

 

Thanks guys for the valuable input on torque / LSD curves / patterns. I feel I can now write a phd thesis on the subject. Lots of technical and valuable knowledge on this forum! Thanks

 

You're right, it would be redundant. Only case I can think of where that might be needed is with negative preload where you need to first overcome preload to have any locking torque from the clutch pack. But I have no idea who uses negative preload and for what, I have only heard of it existing from a paper.