4wd: torque split front-to-rear other than 50/50?

[arturo]

Can torque be split front-to-rear other than 50/50? With the current possibilities for setting up the differentials, I would say no -- but I would be more than happy to be proven wrong!

I am asking because I'd like to make a Peugeot 508 RXH with PSA's Hybrid4 system: a front internal combustion engine (120kW) plus a rear electric engine (27kW).

Ideally, this would be simulated with two separate sources of power, as Project CARS seems to be doing. But I could settle for a central differential if torque could be split so that 81% goes to the front wheels and 19% to the rear wheels.

Now, how could this be implemented? According to the documentation, "if the differential is torque-sensitive, then anti-slip-torque defines the amount of anti-slip torque per input torque". I may be talking nonsense here... but if we could tell the central differential which end is which, then anti-slip-torque = 4.26 will equalize front and rear torques along the desired distribution.

[NaN]

There is a torque_split variable, but it is not exposed, set to 0.5 by default.

[NaN]

I've pushed a patch that allows you to set torque-split value.

[CrystalH]

Technically an electric engine has a different torque curve right? Nearly flat (i.e. independent toque value from rpm).
So probably another instance of CarEngine.
IIRC electric cars don't need a clutch or even a gearbox because of that.
But if it's 120kW and just 27kW then IDK if the difference could be felt in game.
I think I've played with the split value but on gravel, IIRC usuall values close to 0.5 are good, others start to feel just like FWD or RWD.

[arturo]

Technically an electric engine has a different torque curve right? Nearly flat (i.e. independent torque value from rpm).

Again, I may be wrong, but I would say it is the power curve which is nearly flat (which means that torque x rpm is more or less constant).

So probably another instance of CarEngine.

As I said, this would be the ideal solution: when you are flat on the throttle, an additional power input is sent to the rear wheels.

I think I've played with the split value but on gravel, IIRC usuall values close to 0.5 are good, others start to feel just like FWD or RWD.

What is the right 'syntax'? Should I add a torque-split = 0.X line to both the front and the rear differentials, or only to the central one? In the latter case, should the value be .81 or .19?

[NaN]

Code:

Torque split parameter determines the torque split ratio 0.0 to 1.0 between driven axes front and rear (left and right). 0.0 means all torque is applied to front (left).

If this is clear enough, I'll add it to the docs.

[arturo]

A simple test to assess if it is clear enough: if even I can understand it, then everyone else will.
I have to insert, under differential.center, torque-split = 0.19 -- got it right?

[NaN]

A simple test to assess if it is clear enough: if even I can understand it, then everyone else will.
I have to insert, under differential.center, torque-split = 0.19 -- got it right?

Yep, should give you 19% torque at the rear wheels.

[arturo]

Great, I'll start working on it!
For future enchancements, you may want to consider the possibility of a second instance of CarEngine, as CrystalH mentioned, whose power will be liberated when the throttle is open at, say, 90 percent (or, optionally, with a release button). A further step will be to consider the level of energy stored, tipically in a battery or supercapacitor, but also in a flywheel, and how it is recharged through regenerative braking (although this will admittedly add very little to the gameplay experience, and I would be the first to agree that there are more urgent things to do).

[arturo]

By the way, anti-slip is not an optional parameter, as the docs say. It can be set to zero but it has to be present (try to delete it and see what happens).

[arturo]

Maybe I am just too dumb to get the feeling, but I would say that torque-split does no difference at all.
Try driving the TC6 at Australian Countryside. Go to the widened area at the entry of the pits, turn your Traction Control System off and start making circles on the dirt. You cannot make it drift because it is 4wd, right? Now open the .car file and add torque-split = 1.0 under differential.center, hit save and try again. Being in theory a rwd car, it should be very driftable, shouldn't it? Well, it is not. It is exactly the same. And if you change to torque-split = 0.0, in theory a pure fwd vehicle, it is still the same (or at least I feel it so).

[NaN]

I've finally had the time to do some testing.

Running with the TC on Australian Countryside, I've tested with torque-split 1.0 and 0.0 and center diff anti-slip set to 0.0. The difference is clearly there. 0.0 (FWD) is understeering at any time, while 1.0 (RWD) allows induce oversteer by flooring the throttle.

I think the issue is engine power vs tire grip. We might have too much (side) grip, especially on gravel.

[arturo]

First of all, let me thank you for your help and your patience.

I've just understood what my mistake was. When you said "I've pushed a patch that allows you to set torque-split value", I somehow assumed that it was in the latest test build, and I installed it (I was previously using the 'stable' version that Sourceforge pushes by default). Now I see that it was here: https://github.com/VDrift/vdrift/commit/...a5bdf6e4b7. I imagine that I should copy these changes into the files in my /physics folder, but I am a bit afraid of all those plus and minus signs... I think I'd rather wait for the next test build.

I'm very sorry for having wasted your time. You can now guess that I am not an IT guy.

By the way, the .car file for my project is almost ready... At the end, it will be a concept for the future Citroen C5. I already have solid data for engine, transmission, wheels and brakes; for the suspension I will certainly need the help of an expert tuner. But right now my immediate concern is to learn Blender from scratch. That's why I am not in a hurry in waiting for the next test build...

[NaN]

First of all, let me thank you for your help and your patience.

I've just understood what my mistake was. When you said "I've pushed a patch that allows you to set torque-split value", I somehow assumed that it was in the latest test build, and I installed it (I was previously using the 'stable' version that Sourceforge pushes by default). Now I see that it was here: https://github.com/VDrift/vdrift/commit/...a5bdf6e4b7. I imagine that I should copy these changes into the files in my /physics folder, but I am a bit afraid of all those plus and minus signs... I think I'd rather wait for the next test build.

I'm very sorry for having wasted your time. You can now guess that I am not an IT guy.

By the way, the .car file for my project is almost ready... At the end, it will be a concept for the future Citroen C5. I already have solid data for engine, transmission, wheels and brakes; for the suspension I will certainly need the help of an expert tuner. But right now my immediate concern is to learn Blender from scratch. That's why I am not in a hurry in waiting for the next test build...

I see. For some reason I've assumed you would be building vdrift from source.

Get source (you might have to install git first, apt-get ...):

Code:

git clone https://github.com/VDrift/vdrift.git

cd vdrift and pull latest changes:

Code:

git pull

Or get the latest source zip from here: https://github.com/VDrift/vdrift/archive/master.zip

Place/copy data into this vdrift directory (makes things simpler).

Alternatively, I could upload another test build, if you prefer.

[arturo]

I installed git and got the latest code, but either I am doing something wrong or just can't feel the difference. Anyway, as I said, I am currently focused on learning Blender instead of further tweaking the .car file, so I'd rather wait for the next text build just to be sure that the functionality is actually there and then try and get the feeling...