suspension understanding problems

[cologne]

The cars in vDrift should drive in a realistic way, showing the typical caracteristics of the real car. Many of the car setting are irrealistic: F1 with engine in front and 30 cm travel, XS with Cw 0.3 and downforce. Remove all downforce and the car will be uncontrollable!

The F1 should be fast, the XS slower, but controllable drifting and the 360 somewhere inbetween.

I am trying a hundred versions - first the 360 now the XS - .... getting sometimes curious behaviour of the car. Some are driveable, but not good controllable.

I had versions, overturning on brakes. Other "exploded" touching the tracks border. Some had the rear 2 meters in the air, or the front making "wheelies" on small acceleration.

I assumed different suspension-models, but the "practice" doesn´t confirm.

I didn´t really understand, what geometry the suspension-position and hinge settings describe! (And I´m not shure , that the suspension-model is implemented without bugs! Using vdrift_20070327 release, Mac PPC)

????????????????????????

[cotharyus]

Cologne, I agree, there are certainly problems with the suspension implementation. The set up on many of the cars that were done by me were put there to get the most realistic in game performance - some got close, others did not.

The problem is that, in essence, there's only one suspension model in the game. There are obviously many more models in real life.

[reece146]

I believe that the suspension models in VAMOS are not the literal MacPherson Strut, SLA, etc.

They are the resolved, virtual center models. That is why there is only one model.

This is why ALL of the .car file models are wrong compared to the real cars. For a real car most of the suspension hinge points should be outside of the car body, literally.

I don't know this for certain - it is my speculation.

[cologne]

Maybe VAMOS has only one suspension model, but this could be a sufficient approach. Not the best. Good enough to „drift“?

What describe these two parameter: suspension.position, suspension.hinge ?

[joevenzon]

For a real car most of the suspension hinge points should be outside of the car body, literally.

I don't know this for certain - it is my speculation.

I believe you are correct reece.

Here's a previous thread about this:

http://vdrift.net/Forum/viewtopic.php?t=539

Basically, suspension position doesn't mean anything and isn't used for anything. Regarding suspension position, thelusiv wrote up a good summary in the above thread:

Here's how I understand the suspension hinges. The suspension does not follow a straight up-and-down path. Instead it follows a curve. This curve is defined by the hinge point indirectly. It's as if the wheel itself was attached by a straight rod of some kind from the hinge point to the center of the wheel, and the hinge allows up-down movement. Take for example the front wheels, which usually have a hinge point behind them. So if the wheel moves up, since its attached to the hinge behind it, it's also going to move towards the back of the car a little along the curve. The big difficulty in this is predicting how the car will react under this movement.

We should really add some info about this to the wiki.

[thelusiv]

Yeah, that's the most info I know about how suspension works. These guys make good points though, our suspension code needs work, but it won't be easy to fix it. The existing setup approximates an accurate suspension path, but isn't really very accurate in the end. It would be better if we could fully simulate a MacPherson strut or a double-wishbone setup, or both, but the fact remains that will take a lot of work to do...and probably be a lot less efficient than what we have now.

[reece146]

Yeah, that's the most info I know about how suspension works. These guys make good points though, our suspension code needs work, but it won't be easy to fix it. The existing setup approximates an accurate suspension path, but isn't really very accurate in the end. It would be better if we could fully simulate a MacPherson strut or a double-wishbone setup, or both, but the fact remains that will take a lot of work to do...and probably be a lot less efficient than what we have now.

Well, I wouldn't say it is inaccurate, it is only as good as the data put into it.

I'll try to scan and upload some stuff later that shows how to develop the instant center model so we can get closer to reality.

[zimluura]

i think a way to have all the suspension parts drawn, even with gl_lines and/or non-lit primitives, would be a good idea first. then the positioning could be done visually.

after that maybe have ways to temporarily add weight to the car. the equivalents to acceleration transfer and downforce at speed. that way it could easily be seen how the suspension is reacting.

ideally this would look really nice on some test track or in some wind tunnel. but doing the bare minimum it could probably be implemented as diagnostic key commands. actually, is there a command console in vdrift? like in quake? if so, tuning modes could be turned on with commands like:
"tune_speed 120" (km/s)
"tune_acceleration 15" (km/s/s)
"tune_turn 20" (degrees)
or something like that to make all parts on the car act like they're constantly working under those conditions.

[cologne]

The difference between suspension-types are the position of the Instant centers. Some types have chassis-fixed centers, others move the center. This is the difference between vamos and real world, because vamos always uses fixed Instant center. I think, this simple model should be good enough, for playing a game and give the cars a nice behaviour.

To do the things better, we need a better instant center model and not special geometries, simulating real suspensions.

http://en.wikipedia.org/wiki/Suspension_...ant_center

Basically, suspension position doesn't mean anything and isn't used for anything.

This is not true! If suspension-position is before the front-axle, the car goes ... bumping and flying away.

[reece146]

Ok, as an attempt to get everyone on the same page here is a diagram that shows a rough idea of where the instant centre (IC) for a few suspension types end up. Where the lines converge is the IC.


Race and Rally Car Source Book, Allan Staniforth, Haynes Publishing, pp 50

This is the location of the hinge point in the frontal plane.

Now, the wheel is located at the center of the hub of course, this is the length of the lever arm from the hinge or more commonly the front view swing arm (fvsa). I don't know the specifics of how the VAMOS code locates this - is it where the wheel centerline crosses the king pin inclination axis (KPI)?, is it the wheel's contact patch, something else?

To drill down on a SLA setup like the XS uses in real life:


Chassis Engineering, Herb Adams, HPBooks, 1993, pp 47

This is only considering the frontal plane view. The percentage of anti-dive or anti-squat is also related as the geometry of the suspension defines these qualities. This makes the side view of the car also come into play with respect to the location of the hinge point. I believe that in the case of VAMOS the antis are just a number dialed into the .car file and not directly extrapolated from the suspension geometry.

To see what I mean:


Race Car Vehicle Dynamics, Millike & Milliken, SAE Press, 1995, pp 632

The big problem with this model as I see it is that it is only close to realistic values within small suspension deflections. Not a big deal and arguable that it makes the sim more playable as there is only so much feedback you can get from the console even with working force feed controls.

We haven't really touched scrub radius or KPI at this point.

Btw, I have a MY2000 "XS" in my workshop. I can try to get real world numbers off it but being put away for the winter there is a lot of crap in the way. I can try to get at it over the next few weeks.

[thelusiv]

Very nice diagrams, Reece! We need all this stuff on the wiki! It could be developed into an article explaining how to use Blender to find coordinates for a given car and how to put that into the .car file to make it work in VDrift.

[reece146]

It was my intent to do something for the wiki but then the PPC port seemed to go wonky for a while and I kinda lost interest with other stuff getting priority over the summer (racing!). I haven't been active here for a while. Is PPC working now?

FWIW, this is what the SLA on an XS is supposed to look like. First front:



and rear:

[cotharyus]

reece, I just put a new Mac build on my website today - it *should* work - please test it and let me know. It's in the downloads section at http://cotharyus.net

My take on all of this is I have looked at the suspension settings in the .car files to see what can be adjusted and tweaked, and it looks pretty limited. In most cases, you have to do the best you can with that you have. In some cases, it's not too bad, in other cases, it can be had to get exactly what you're after.

[reece146]

reece, I just put a new Mac build on my website today - it *should* work - please test it and let me know. It's in the downloads section at http://cotharyus.net

Code:

Exception:  EXC_BAD_ACCESS (0x0001)
Codes:      KERN_PROTECTION_FAILURE (0x0002) at 0x00000000

Is there any point in me opening a thread for this? We shouldn't discuss here. Back channel PM me if you want.

[thelusiv]

Code:

Exception:  EXC_BAD_ACCESS (0x0001)
Codes:      KERN_PROTECTION_FAILURE (0x0002) at 0x00000000

Is there any point in me opening a thread for this? We shouldn't discuss here. Back channel PM me if you want.

Yeah, if you're willing to do some testing. Everyone messing with the Mac build doesn't have a PPC machine anymore. If you're willing to test, that could be very beneficial to there being a Mac PPC version working.