Updates have been made! See edited first post!
I’ve obtained a speed time curve:
On a default baseplate:
On the Test Track
The car seems to bounce around lots on the test track but not on the default baseplate, which is odd as they should have the same Elasticity.
Edit:
This is in MPH against time in seconds. The 0-60 is incredibly fast, yet the car can barely break 100, and even my little compact can make it to that speed. I’m not introducing any frictional forces myself in this test, no air resistance, so I’m sort of at a loss of what to do here.
How are you generating power? Are you using one of the bodymovers?
If so that could account for some of the issues; I’ve tried to use them for vehicles before and found them rather unreliable, and more than a tad difficult to try convert the units.
The curve follows a fairly standard shape, but the reason the first region is so fast is due to the physics applying the max torque instantly. If you set up a table of torque figures based on vehicle speed then it should show a more natural curve. Also, as I guess you don’t have a clutch modelled (yet… ;)) ideally the torque should have a base value of ~80 (Nm) between 0 and 10mph.
i.e.
torque = {
s0 = 80, --*10Nm
s10 = 10,
s20 = 23,
s30 = 30,
s40 = 33,
s50 = 32,
s60 = 31,
s70 = 29,
s80 = 26,
s90 = 23
}
while wait() do
local speed = math.ceil(vehSpeed/10)*10
local speedL = math.floor(vehSpeed/10)*10
local upper = torque["s"..speed]
local lower = torque["s"..speed-10]
vehicle.MotorTorque = (lower+((vehSpeed-speedL)/10)*(upper-lower))*10 -- Interp the torque points
end
EDIT: I’m aware this is totally the wrong way around to do the torque calculations but it kinda works lel
1 Like
The wheels have two parts with BodyThrust in them, which generate the torque. Is there an equation I can use that links the torque sent to the wheels, the engine RPM, and vehicle speed?
Edit: Found the problem. This is a graph of WheelRPM against time with a constant torque applied and all collisions off
It should be directly proportional, but instead there’s a huge friction being applied despite no collisions and so nothing to generate friction with.
Power = EngineSpeedInRadiansPerSecond*CurrentTorque
To get rid of what appears to be wheel inertia (?) just make the density 0.01 and make a part in the chassis somewhere to add the extra weight. Maybe that’d work? What kind of joint are you using? It might be a good idea to switch to the new HingeConstraint (I think you mentioned above that it’s just a standard old hinge?)
Also, unless you already have, I’d suggest giving BodyAngularVelocity a try… maybe that’d solve a few issues or at least make it more straightforward.
BTW, I realised it’s probably better to interpolate the motor torque against engine speed (rpm), not the vehicle speed lol.
EDIT: how the heck did I reply to myself…
1 Like
engineSpeed = wheelSpeed*gearRatio
torque = TorqueCurve(engineSpeed)*gearRatio
more complicated on auto because torque converter.
Torsion via BodyThrusts works well for me, so it’s probably not the cause of the issue. If the wheel has a high moment of inertia it wouldn’t cause drag and would probably keep the car in motion (if anything).
Yeah, HingeConstraints will be ideal when they’re released.
That’s what I use currently
But wouldn’t this just result in even less torque at high speed, making the problem worse?
I think I’m just going to wait for new constraints then start again from there.
Oh, that’s probably your issue. Glue tends to work better than hinges atm.
“Less torque at high speed” is perfectly natural, but tbh that equation is pretty much useless taken out of context. Probably better to ignore it lel
Yeah, they don’t wobble much which is great, but they seem to experience absurdly high friction, which I think is why the acceleration from 0-60 is F1 speed yet the car is incapable of breaking more than about 120.
We haven’t encountered those sorts of issues at Suspension Test, which is why I’m dubious that it’s the fault of physics. You can try printing the torque that’s being delivered to the wheels every frame and see where that takes you.
at least now you have a good electric car.
The graph I posted earlier is just a wheel on a hinge, with two parts with body thrust in welded to it. The body thrusts are constant, all friction is 0, all collisions are off. It should be directly proportional, a=F/m , f and m are constant, but it tapers off quickly, showing some sort of string frictional force.
Hinges are unstable at high speeds and torques. Have you done the same test with glue?
Am I doing something wrong? It wont move, and if I ramp the power up it still wont move until the wheel flys of completely.
Try messing around with the glue’s face values.
I’ve added an ABS system for the brakes, updated the gear ratios, and fiddled with the power and friction. I’ve tried to counter the weird extra friction in the hinges with an extra force, which mostly works but needs a little extra tuning. The cars now do 0-60 in 8 seconds and max out at 120, instead of 0-60 in 2 seconds and max out at 100, which only RallyX cars do.
However the cars don’t have enough torque, tractions fine, to get up anything other than a shallow hill. I’m thinking it’s because I’m treating my cars as full size but the default roblox gravity is for cars measured in cm not m?
@NWSpacek @Fractality_alt
Did some tests and cylinders are definitely not real cylinders: At about 120MPH cylinder wheeled cars will start to jump around uncontrollably, but using ball wheels results in a perfectly smooth ride at all speeds.
3 Likes
that’s problematic. would you mind filing a bug report?
I’m guessing its intended behaviour though: It’s as NWSpaceK said, it’s a prism and not a cylinder.
I’ve switched to a sphere wheeled car, but this introduces a new problem: As the wheels tilt with the chassis in turning, the part of the sphere in contact with the road changes, and the radius of the rotation of the wheel effectively becomes smaller. This slows the wheels on one side and speeds the wheels on the other, making the car enter uncontrollable turns when using soft and high suspensions.