Cylinders are still affected by high instability when rotating at high speed, they hop quite often and sometimes it even makes my vehicles jump.
It can be recreated by just going fast (over 80 sps) on a cylinder wheels equipped vehicle.
https://gyazo.com/d95c7fd437b23770aceef6ed9fc60da1
(before you wonder, i cant use balls because im replicating tyre profiles for motorcycles!)
There’s a new physics update just around the corner that makes wheels more stable at high velocities by using less forces to produce the same rotation speeds. This might solve that. Can you access gametest3 and try this on that build?
is this PGS or spring?
Can you reproduce this same jumping on a simple model consisting of a body and 4 wheels with 4 motors? I’d like to eliminate the possibility that the jumping is due to some weird scripting issues.
If you do reproduce it, please provide the simpler model here.
Seems like the issue is when cylinders switch the collision from a brick to another brick.
Here it is a perfectly flat straight composed by multiple bricks, everytime there is a brick change the cylinder gets unstable and hops.
Just press play with character, everything is already set up
Cylinder problem.rbxl (16.8 KB)
I seem to remember a discussion some time ago about a problem with trains using cylinders ‘hopping’ when they went from rail to rail.
I can’t find that thread now, but the problem was explained to be that the ‘wheel’ would roll off of the previous rail part and start to descend before colliding with the second part, even if the parts were flush against one another. Once it collided with the second part it would have already penetrated a fraction down into it, causing the physics solver to ‘kick’ the part back out and creating an upward force.
sounds legit, but why doesnt it happend with spheres?
anyways i hope there is a fix for this, spheres dont seem to simulate centripetal forces to make motorbikes steer by just leaning.
its PGS
I was the one who brought that up 
Thanks for the repro. Cylinders are bit more complicated than spheres for collision detection and do experience imperfect collisions on plate boundaries. There are mitigation steps you can take to reduce this effect:
Cylinder problem mitigation.rbxl (18.6 KB)
I have a working example of a vehicle that can reach high speeds and doesn’t experience much of this effect:
Is this using the new update or is that not out yet?
Edit: Oh god. No it does not use the new update. I have to get the GT3 studio for that.
The new update won’t affect this. This is problem is about collision detection, where the angular damping update is about torque.
These are unrelated.
i cant really make the wheels bigger or increase the friction, elasticity is already set to 0 and i’ve added a suspension system, which gets rid of most small ones but bigger jumps still occur
https://gyazo.com/7fb1d3255a4f64e8929ead60cefb825d
I don’t have time to keep experimenting with it. Overall the spheres are indeed way smoother. They will probably interfere with 2 wheeled physics but worth a try to me if i was working on that. It seems like they might be really good for 4 wheels. When the wheels are free-wheeling they get wobbly (too many rpms and sphere physics getting crazy maybe?). Most likely the wobbling can be fixed by experimenting with the motor parameters. The karts seem to drive normal. Without adding controls for further testing I can’t say.
as long as the physics behavior stays constant then one possible cover-up for the wobbling could be to attach the sphere and wheel to separate motors with the wheel scripted to match the speed of the sphere.
cylinder sphere comparison.rbxl (566.7 KB)
Its not like i havent tried spheres
spheres have been in my old bike chassis for 2-3 years, its just that they are not good for it
i still use spheres for cars, since there’s no need for cylinders
Make the body of the vehicle heavier than it is, and adjust the springs accordingly. Also the knuckle (attachment of the wheel) must be heavy otherwise you’ll get vibrations.