rec.autos.simulators

Here's a nice one that some people stumbled upon on the Racer forum,
and frankly I'm not sure what to do with it:

(see also
http://www.racesimcentral.net/)

The PDF that is talked about is a standard talk about the Magic
Formula. The problem is about the sign of camber.

Quote from the forum:
---
The camber sign is related to which side of the car the tire is. When
camber is negative the tire leans in. In fabulous ASCII art:

/ - \ (looking from the rear of the car forward to the nose)

Both tires have negative camber. However, Fy (lateral force) points in
the same direction for both tires, as it is a world axis (!). The sign
of gamma (camber) in the formula in that PDF therefore is undefined?!

The reason being that for both tires in the above diagram, you'd
expect the delta-Fy to be negative for one tire, and positive for the
other. But from the formula the camber (gamma) effect is the same for
both tires (on Fy). This clearly isn't right.

For this camber-dependeny reason, there is inclination angle, which is
the angle as viewed from the rear of the vehicle. So for the above
diagram, the left tire has a negative incl. angle, the right tire has
a positive incl. angle. This is what is used in Racer.

So the problem is: in the PDF and the Pacejka model, what is the
(world) direction of positive camber?? With just 'camber', you'd
expect different formulae for the left and right side tires of the
car! (sign changes)

In other words, is the Pacejka tire model completely broke?  Not
really ofcourse, but I have no idea on what side the Pacejka model is
based.
---

Anyone?

Ruud van Gaal
Free car sim: http://www.racesimcentral.net/
Pencil art  : http://www.racesimcentral.net/

[big snip..]

To start the discussion going... I'll say... umm... the left side?
;-)

itazura

I saw that thread :-)

To fix it up, either negative or positive inclination is going to increase
force along the positive direction of your lateral axis.  Couldn't you try both
and see which it is? Imagine looking down on one tire with no car attached
instead...  The "right" or spin axis vector is either pointing up or down in
the presence of camber.  You could define either one (up or down) as being
negative inclination depending on how your coordinate systems work.  I'd try
both and get rid of camber entirely in the physics model and just use the angle
between the spin axis vector and surface normal vector (minus 90 degrees) to
get inclination, then try both signs in Pacejka and see which works correctly.

Todd Wasson
---
Performance Simulations
Drag Racing and Top Speed Prediction
Software
http://PerformanceSimulations.Com

My little car sim screenshots:
http://performancesimulations.com/scnshot4.htm

Whoops, one more thing.  After you've got that worked out and you still want to
display camber on screen, it's easiest (for me anyway) to just make that a text
operation where you reverse the signs for one side of the car.  You're already
doing that though, I'm sure :-)

As far as the other part goes, what I meant was if your positive lateral
direction is to the right, then find out which sign when fed into the Magic
model will result in increased force to the right (most likely negative
inclination).  If your positive axis goes left instead, swap signs and you
should be fine I think.

Todd Wasson
---
Performance Simulations
Drag Racing and Top Speed Prediction
Software
http://PerformanceSimulations.Com

My little car sim screenshots:
http://performancesimulations.com/scnshot4.htm

Cheers,
Steve.

I think something like this is most traditional ;)

- Matt

LOL !  I didn't guys like you were prone to this sort of thing too , Matt! :-D

Thanks for the chuckle.  You wouldn't even believe the blunder I'm trying to
fix now :-P

Todd Wasson
---
Performance Simulations
Drag Racing and Top Speed Prediction
Software
http://PerformanceSimulations.Com

My little car sim screenshots:
http://performancesimulations.com/scnshot4.htm

Sounds like you are not using SAE axis system (no reason why
you should be)...  RCVD pages 62-63.

We actually went through and called it inclination angle everywhere
that we use it (at least in all our docs, not sure about comments
all through the code).

And if you choose a non-standard coordinate system you
need even more comments. Starboard is the same direction
on every ship for a reason :)

I'm working on someone elses code now where whole
classes are mysterious. This thing isn't working
because the "pool" member variable is empty. The purpose
of this pool object could have been written in a sentence
in 5 seconds and I would have then immediately known that
not having one was a serious problem. Instead I spent
at lest an hour figuring out what it was, what object
was supposed to have created it and why it failed.

I believe people sometimes think that those type of comments
aren't needed if they are the only person that will ever
be looking at this code.  Until 6 months later and they
can't remember what the class is for either! :)

- Matt

when doing "random sign flipping", you can often get something where two
things are backwards, but (at least mostly) cancel each other out, but
then when you make some other change, things will be off in some way...

Not that I always follow my own advice, but at least I feel guilty when
I do...   ;)  and try to remember to go through the code again to
double-check.  (It's very easy to get impatient when you have a model
that is very close to working and you're just itching to be able to
drive it around and do donuts and stuff instead of squinting at rows of
numbers...)
</pedantic>

-Dave P.
MGI

PS.  Hey Doug, I think RCVD could use another chapter on tire
behavior--it would be great to have one overview reference that went
over more tire modelling stuff, hopefully all using the same notation
conventions... ;-)

[snip]

<snip>

I'm somewhat familiar with this occurance :-D

My vehicle model runs in one axis system, my graphics testbed (OpenGL) runs in
another, and a game this is going into runs in still another (DirectX).  I have
had at least a couple of funny - signs scattered about to get things working
right on all systems at the same time :-D

Oh well, some of us have to learn everything the hard way :-D

Todd Wasson
---
Performance Simulations
Drag Racing and Top Speed Prediction
Software
http://PerformanceSimulations.Com

My little car sim screenshots:
http://performancesimulations.com/scnshot4.htm

If the game is still in its infancy, and the programmer is not too
religious; you can switch DirectX into a righthanded system (making it
the same as OpenGL).

If nobody had invented the minus sign, we'd sure be up a smelly brown
creek yes. ;-)

Dave is right though; 'debug' minus signs have a high probability of
coming back and haunting you on the search for very deeply hidden
bugs.

Ruud van Gaal
Free car sim: http://www.racer.nl/
Pencil art  : http://www.marketgraph.nl/gallery/

Ah well, that's his problem ;-)  In all seriousness, it's all sorted out fine
now.  

Yep :-)

Todd Wasson
---
Performance Simulations
Drag Racing and Top Speed Prediction
Software
http://PerformanceSimulations.Com

My little car sim screenshots:
http://performancesimulations.com/scnshot4.htm

Enough perhaps at times to get you in a straightjacket if you don't go
out once in a while. ;-)

So I guess you do all your debugging output in SI units?

Well, you can call that a 'feature', I'd say. :) Call it 'Old American
Cars Heat' and you're set...

Ruud van Gaal
Free car sim: http://www.racer.nl/
Pencil art  : http://www.marketgraph.nl/gallery/