Car physics; steering caster angle

Hi all, I'm trying to get a feel for suspension and steering systems, but I don't have a good starting point yet (mathematically). I know what the caster angle is, have looked at the pretty pictures in RCVD explaining the dozen or so terms that describe a steering axis angles and such, but I can't put this into mathematics. A concrete example, if I have a certain caster angle, what would this do (mathematically) with the forces both in the tire itself and in the steering feedback? I've got the impression that caster angle should make a lot of difference in the driving behavior at high speed (I've got aerodynamics more or less in, so ouch I'll have to drive an F1 car soon), but I wouldn't know how this behavior changes (also because I've never 'felt' in real life what the difference between different caster angles is). Thanks for any pointers on this one, Ruud van Gaal, GPL Rank +53.25 Pencil art : http://www.marketgraph.nl/gallery/ Free car sim : http://www.marketgraph.nl/gallery/racer/

from old family knowledge. Increasing use of Castor(oil) tends to make the rear end loose and unstable, loosing control at inopportune times can be the result of too much Castor. dave henrie "Ruud van Gaal" <[email hidden]> wrote in message news:[email hidden]... > Hi all, > > I'm trying to get a feel for suspension and steering systems, butI...(snip)

Hi Ruud, first of all, let's limit the discussion to small steering/caster angles, which is what is typically happening in the real life as well. The forces on the tyres themselves should remain unchanged. That is to say, the Pacejka formulas that you use aren't changed at all. However, the caster affects how much additional camber you get when you turn the wheel, and this is reflected in the forces on the tyres. A linear approximation (correct up to the sign) would be AdditionalCamber[rad] = Caster[rad] * WheelRotation[rad]. You should actually use Sin(Caster) (I think :) ), but since we'll be dealing with small angles, the above should suffice. For steering feedback, what caster does is that it throws the alignment of the rotation of the wheel around its steering axis with the contact patch off. At this point not only the aligning torque contribuites to steering feedback, but the lateral force as well. The linear approximation in caster angle for the total torque exerted around the steering axis is M_total = M_aligning + Caster[rad] * R * F_lateral, which is accurate up to the sign again ;). Use this instead of only the M_aligning for force feedback. The way how I do it all, though, is to define suspension in terms of 5 links (of which one can be movable for steering), and calculate the force on the steering link to get force feedback. That way I get all of the above for free (admittedly, after the pain of making this all work ;) ). -Gregor Ruud van Gaal wrote:

Quoted text (24 lines)

> > Hi all, > > I'm trying to get a feel for suspension and steering systems, but I > don't have a good starting point yet (mathematically). > I know what the caster angle is, have looked at the pretty pictures in > RCVD explaining the dozen or so terms that describe a steering axis > angles and such, but I can't put this into mathematics. > > A concrete example, if I have a certain caster angle, what would this > do (mathematically) with the forces both in the tire itself and in the > steering feedback? > I've got the impression that caster angle should make a lot of > difference in the driving behavior at high speed (I've got > aerodynamics more or less in, so ouch I'll have to drive an F1 car > soon), but I wouldn't know how this behavior changes (also because > I've never 'felt' in real life what the difference between different > caster angles is). > > Thanks for any pointers on this one, > > Ruud van Gaal, GPL Rank +53.25 > Pencil art : http://www.marketgraph.nl/gallery/ > Free car sim : http://www.marketgraph.nl/gallery/racer/

> I'm trying to get a feel for suspension and steering systems, but I > don't have a good starting point yet (mathematically). If it's not answered in this series, I'm sure the author knows.... http://bovineracing.com/EOA/pt1/Pt1.htm

I'm just now getting around to mastering FF. Unlike every other FF-equipped racing sim, GPL models caster return. Thus, it's a remarkably effective teaching tool. If you're driving correctly (at least, if your steering input is correct), you'll never feel the caster return. If you're doing it wrong, you'll feel it: the c.r. practically rips the wheel out of your hands. [GPL's caster return also provides some otherwise-hidden input on how your shocks are set up. If you don't have the suspension properly damped, the caster return will oscillate, which you can definitely feel thru the wheel.] Remarkable. We're still discovering new stuff in GPL 3 years on. --Steve "Ruud van Gaal" <[email hidden]> wrote in message news:[email hidden]...

Quoted text (23 lines)

> Hi all, > > I'm trying to get a feel for suspension and steering systems, but I > don't have a good starting point yet (mathematically). > I know what the caster angle is, have looked at the pretty pictures in > RCVD explaining the dozen or so terms that describe a steering axis > angles and such, but I can't put this into mathematics. > > A concrete example, if I have a certain caster angle, what would this > do (mathematically) with the forces both in the tire itself and in the > steering feedback? > I've got the impression that caster angle should make a lot of > difference in the driving behavior at high speed (I've got > aerodynamics more or less in, so ouch I'll have to drive an F1 car > soon), but I wouldn't know how this behavior changes (also because > I've never 'felt' in real life what the difference between different > caster angles is). > > Thanks for any pointers on this one, > > Ruud van Gaal, GPL Rank +53.25 > Pencil art : http://www.marketgraph.nl/gallery/ > Free car sim : http://www.marketgraph.nl/gallery/racer/

On Mon, 23 Jul 2001 13:08:15 -0400, "Art McEwen" <[email hidden]> wrote: >> I'm trying to get a feel for suspension and steering systems, but I >> don't have a good starting point yet (mathematically). > >If it's not answered in this series, I'm sure the author knows.... > >http://bovineracing.com/EOA/pt1/Pt1.htm Ah, that one shows up in every place again and again! ;-) Ruud van Gaal, GPL Rank +53.25 Pencil art : http://www.marketgraph.nl/gallery/ Free car sim : http://www.marketgraph.nl/gallery/racer/

On Mon, 23 Jul 2001 16:49:51 GMT, "Dave Henrie" <[email hidden]> wrote: > from old family knowledge. >Increasing use of Castor(oil) tends to make the rear end loose and unstable, >loosing control at inopportune times can be the result of too much Castor. Ah, and what about Pollux? Ruud van Gaal, GPL Rank +53.25 Pencil art : http://www.marketgraph.nl/gallery/ Free car sim : http://www.marketgraph.nl/gallery/racer/

On Mon, 23 Jul 2001 17:56:00 GMT, "Stephen Smith" <[email hidden]> wrote:

Quoted text (10 lines)

>I'm just now getting around to mastering FF. Unlike every other FF-equipped >racing sim, GPL models caster return. Thus, it's a remarkably effective >teaching tool. If you're driving correctly (at least, if your steering >input is correct), you'll never feel the caster return. If you're doing it >wrong, you'll feel it: the c.r. practically rips the wheel out of your >hands. > >[GPL's caster return also provides some otherwise-hidden input on how your >shocks are set up. If you don't have the suspension properly damped, the >caster return will oscillate, which you can definitely feel thru the wheel.]

Ah, I always thought the oscillations were caused by the delay set too high or low in the FF section. Now that I reread the previous paragraph, no I never felt anything remotely like it! ;-) Uh... >Remarkable. We're still discovering new stuff in GPL 3 years on. It's amazing at the complexity that complex systems give; you can never really understand the results your own product somehow, but it's there. :) Ruud van Gaal, GPL Rank +53.25 Pencil art : http://www.marketgraph.nl/gallery/ Free car sim : http://www.marketgraph.nl/gallery/racer/

>For steering feedback, what caster does is that it throws the alignment >of the rotation of the wheel around its steering axis with the contact >patch off. At this point not only the aligning torque contribuites to >steering feedback, but the lateral force as well. I thought lateral force caused the aligning torque to begin with. ?

Quoted text (8 lines)

>The way how I do it all, though, is to define suspension in terms of 5 >links (of which one can be movable for steering), and calculate the >force on the steering link to get force feedback. That way I get all of >the above for free (admittedly, after the pain of making this all work >;) ). > >-Gregor >

Are you calculating forces independently at all five links? I've currently got a 2-D approximation (front view only) of the upper and lower control arms and the shock/spring axis so bump camber occurs and dynamic motion travel and shock/spring angle effect wheel rate. My plan was to rotate the tire orientation vector and contact patch force centroid around the steering axis to get wheel rise and fall with steering, as well as camber change from caster and kingpin angles, run the lateral force (in body coords) along the ground plane, then apply the jacking force from roll center location as well. If this was done in the side view (instant center) and front view (roll center), would it have the same effect as calculating and applying the forces through all the links individually? It seems this is what RCVD describes. Thanks, Todd Wasson --- Performance Simulations Drag Racing and Top Speed Prediction Software http://PerformanceSimulations.Com

Hi Gregor,

Quoted text (8 lines)

>For steering feedback, what caster does is that it throws the alignment >of the rotation of the wheel around its steering axis with the contact >patch off. At this point not only the aligning torque contribuites to >steering feedback, but the lateral force as well. The linear >approximation in caster angle for the total torque exerted around the >steering axis is > >M_total = M_aligning + Caster[rad] * R * F_lateral,

And here R is the radius of the normal contact patch center to the point where the new axis (with caster) hits the ground? Sounds like it only gives more feedback than usual, but doesn't really make that much of a difference, or does it? Ruud van Gaal, GPL Rank +53.25 Pencil art : http://www.marketgraph.nl/gallery/ Free car sim : http://www.marketgraph.nl/gallery/racer/

<<Ah, I always thought the oscillations were caused by the delay set too high or low in the FF section. Now that I reread the previous paragraph, no I never felt anything remotely like it!>> Check it out: run the FF values (particularly damping) up & down; say 0 to 80; with yer stock setup. Next, leave the damping where you usually leave it (I have it at 30 in both the LWFF Profiler & core.ini; I'm not sure if they're additive, mutually exclusive, or what), and run the springs & shocks up & down (1/1 & 50 lb. springs & bars, to 5/5 & 200 lbs. springs & bars, with GRE). See? Dead soft, what yer feeling is the chassis rocking from side to side, and, as it does so, the see-sawing tries to steer the car, and you get a "tank-slapper." Damp the rocking and you damp the caster kickback. Is this a great game or what? --Steve

Interesting....I had trouble with tank slappers when I was running very soft and high Brabham setups. Ben Stephen Smith wrote in message <8p477.106870$[email hidden]>...

Quoted text (8 lines)

><<Ah, I always thought the oscillations were caused by the delay set too >high or low in the FF section. Now that I reread the previous paragraph, no >I never felt anything remotely like it!>> > >Check it out: run the FF values (particularly damping) up & down; say 0 to >80; with yer stock setup. Next, leave the damping where you usually leave >it (I have it at 30 in both the LWFF Profiler & core.ini; I'm not sure if >they're additive, mutually exclusive, or what), and run the springs &

shocks >up & down (1/1 & 50 lb. springs & bars, to 5/5 & 200 lbs. springs & bars, >with GRE). See? Dead soft, what yer feeling is the chassis rocking from >side to side, and, as it does so, the see-sawing tries to steer the car,and

Quoted text (8 lines)

>you get a "tank-slapper." Damp the rocking and you damp the caster >kickback. > >Is this a great game or what? > >--Steve > >

Hi Ruud, Ruud van Gaal wrote:

Quoted text (7 lines)

> > And here R is the radius of the normal contact patch center to the > point where the new axis (with caster) hits the ground? > > Sounds like it only gives more feedback than usual, but doesn't really > make that much of a difference, or does it? >

Well, if you need to define R this precisely, then the R is actually the length of projection of the vector from the center of the wheel to the contact patch onto the new steering axis. But since we are in a linear approximation here anyway, setting it to the radius of the tyre is what is sufficient. Indeed, it's mostly about the feedback, and the camber behaviour in steering inputs. Definitely a secondary thing to consider in a car sim. If you want to see the effects in a sim, Nascar 4 does this extremely well. It makes a huge difference in FF and control of the vehicle, but I couldn't notice any performance differences due to additional camber, at least not with my untidy driving style ;). -Gregor

Hi Todd, "J. Todd Wasson" wrote:

Quoted text (8 lines)

> > >For steering feedback, what caster does is that it throws the alignment > >of the rotation of the wheel around its steering axis with the contact > >patch off. At this point not only the aligning torque contribuites to > >steering feedback, but the lateral force as well. > > I thought lateral force caused the aligning torque to begin with. ? >

Indeed, that is correct. In my sim, for example, I vary the position at which the lateral force applies, I do not have an actual aligning torque variable. It's only a matter of description, though. You can always consider a force acting at a distance from the axis, or put that force onto the axis and introduce an additional torque, the effect on a rigid body is always the same.

Quoted text (13 lines)

> > Are you calculating forces independently at all five links? I've currently > got a 2-D approximation (front view only) of the upper and lower control arms > and the shock/spring axis so bump camber occurs and dynamic motion travel and > shock/spring angle effect wheel rate. My plan was to rotate the tire > orientation vector and contact patch force centroid around the steering axis to > get wheel rise and fall with steering, as well as camber change from caster and > kingpin angles, run the lateral force (in body coords) along the ground plane, > then apply the jacking force from roll center location as well. If this was > done in the side view (instant center) and front view (roll center), would it > have the same effect as calculating and applying the forces through all the > links individually? It seems this is what RCVD describes. >

Yes, I am calculating individual forces at all five (six, if you consider the spring as an additional link) links. It's a matter of solving a nonlinear system of equations iteratively (3-5 steps each iteration). Also, bear in mind that at this point I still neglect the mass of the tyres and with that the high frequency wheel hopping mode. Since I am optimizing it for an aero formula kind of a sim, this wheel hopping mode is sent to such high frequencies and is usually well damped, so I hope it's not too bad of an approximation. If what you describe was done, you'd indeed end up with a very similar system, as the wheel motion and caster/kingpin descriptions can be largely separated, at least at not too large wheel travels. You only need to keep in mind that all of the descriptions of these axes, along with roll centers etc. are linear approximations for small wheel travel, which, however should usually be more than sufficient and is a complete description for such travels. -Gregor

On Tue, 24 Jul 2001 10:38:54 +0200, Gregor Veble <[email hidden]> wrote:

Quoted text (15 lines)

>Hi Ruud, > >Ruud van Gaal wrote: >> >> And here R is the radius of the normal contact patch center to the >> point where the new axis (with caster) hits the ground? >> >> Sounds like it only gives more feedback than usual, but doesn't really >> make that much of a difference, or does it? > >Well, if you need to define R this precisely, then the R is actually the >length of projection of the vector from the center of the wheel to the >contact patch onto the new steering axis. But since we are in a linear >approximation here anyway, setting it to the radius of the tyre is what >is sufficient.

Ah, so that would be cos(casterAngle)*tireRadius. No approximation needed probably, since this can be precalculated. Unless ofcourse you build in steering compliance. Skip that, I'll just precompute it and keep it static. ;-) >Indeed, it's mostly about the feedback, and the camber behaviour in >steering inputs. Definitely a secondary thing to consider in a car sim. >If you want to see the effects in a sim, Nascar 4 does this extremely >well. It makes a huge difference in FF and control of the vehicle, but I >couldn't notice any performance differences due to additional camber, at >least not with my untidy driving style ;). Sounds good, I don't want too many primary driving behavior anymore! Still, it seems nice to build in, since I noticed that the force feedback is very important to feel your limits, especially now that my skid volume code is in bad need of repairs! ;-) BTW I think that: additionalCamber=caster*wheelRotation should be additionalCamber*sin(wheelRotation), not sin(caster). But I may be wrong here. :) And besides, perhaps it's so close to linear for normal steering that the sin() isn't worth the calculation. Getting some optimization vibes up my spines. Thanks, Ruud van Gaal, GPL Rank +53.25 Pencil art : http://www.marketgraph.nl/gallery/ Free car sim : http://www.marketgraph.nl/gallery/racer/

Hi Ruud, the pain of thinking about combined full 3D rotations is too hard for myself to get into right now during these hot summer days :). That's why everything I've written is a linear approximation, where rotations can be done interchangeably. In all seriousness, when angles involved are lower than 0.1 radians (that's about 6 degrees), if taking 1 instead of cos(x) and x instead of sin(x), you are making an error of about 0.5 percent, which is insignificant. Of course, all the formulas can be given in an exact way and are numercially relativelty easy to do, but would bring very little in accuracy and a lot in headaches to get them absolutely right ;). -Gregor Ruud van Gaal wrote:

Quoted text (25 lines)

> > Ah, so that would be cos(casterAngle)*tireRadius. No approximation > needed probably, since this can be precalculated. Unless ofcourse you > build in steering compliance. Skip that, I'll just precompute it and > keep it static. ;-) > > >Indeed, it's mostly about the feedback, and the camber behaviour in > >steering inputs. Definitely a secondary thing to consider in a car sim. > >If you want to see the effects in a sim, Nascar 4 does this extremely > >well. It makes a huge difference in FF and control of the vehicle, but I > >couldn't notice any performance differences due to additional camber, at > >least not with my untidy driving style ;). > > Sounds good, I don't want too many primary driving behavior anymore! > Still, it seems nice to build in, since I noticed that the force > feedback is very important to feel your limits, especially now that my > skid volume code is in bad need of repairs! ;-) > > BTW I think that: > additionalCamber=caster*wheelRotation should be > additionalCamber*sin(wheelRotation), not sin(caster). But I may be > wrong here. :) And besides, perhaps it's so close to linear for normal > steering that the sin() isn't worth the calculation. Getting some > optimization vibes up my spines. >

On Tue, 24 Jul 2001 12:32:02 +0200, Gregor Veble <[email hidden]> wrote: >the pain of thinking about combined full 3D rotations is too hard for >myself to get into right now during these hot summer days :). Indeed, I should go out and eat lunch in the sun right now! And you're probably right, when nitpicking about sin(a) instead of a, I should then immediately take the full 3D steering angles and then just give up and eat a sandwich and enjoy the hot sun. >In all seriousness, when angles involved are >lower than 0.1 radians (that's about 6 degrees), if taking 1 instead of >cos(x) and x instead of sin(x), you are making an error of about 0.5 >percent, which is insignificant. Ofcourse, I must sometimes remind myself I'm making a game here, which is always off by more than the 0.5% probably. ;-) Ruud van Gaal, GPL Rank +53.25 Pencil art : http://www.marketgraph.nl/gallery/ Free car sim : http://www.marketgraph.nl/gallery/racer/

Quoted text (16 lines)

>> >For steering feedback, what caster does is that it throws the alignment >> >of the rotation of the wheel around its steering axis with the contact >> >patch off. At this point not only the aligning torque contribuites to >> >steering feedback, but the lateral force as well. >> >> I thought lateral force caused the aligning torque to begin with. ? >> > >Indeed, that is correct. In my sim, for example, I vary the position at >which the lateral force applies, I do not have an actual aligning torque >variable. It's only a matter of description, though. You can always >consider a force acting at a distance from the axis, or put that force >onto the axis and introduce an additional torque, the effect on a rigid >body is always the same. > >

Ok, just making sure I hadn't missed something :-)

Quoted text (39 lines)

>> Are you calculating forces independently at all five links? I've >currently >> got a 2-D approximation (front view only) of the upper and lower control >arms >> and the shock/spring axis so bump camber occurs and dynamic motion travel >and >> shock/spring angle effect wheel rate. My plan was to rotate the tire >> orientation vector and contact patch force centroid around the steering >axis to >> get wheel rise and fall with steering, as well as camber change from caster >and >> kingpin angles, run the lateral force (in body coords) along the ground >plane, >> then apply the jacking force from roll center location as well. If this >was >> done in the side view (instant center) and front view (roll center), would >it >> have the same effect as calculating and applying the forces through all the >> links individually? It seems this is what RCVD describes. >> > >Yes, I am calculating individual forces at all five (six, if you >consider the spring as an additional link) links. It's a matter of >solving a nonlinear system of equations iteratively (3-5 steps each >iteration). Also, bear in mind that at this point I still neglect the >mass of the tyres and with that the high frequency wheel hopping mode. >Since I am optimizing it for an aero formula kind of a sim, this wheel >hopping mode is sent to such high frequencies and is usually well >damped, so I hope it's not too bad of an approximation. > >If what you describe was done, you'd indeed end up with a very similar >system, as the wheel motion and caster/kingpin descriptions can be >largely separated, at least at not too large wheel travels. You only >need to keep in mind that all of the descriptions of these axes, along >with roll centers etc. are linear approximations for small wheel travel, >which, however should usually be more than sufficient and is a complete >description for such travels. > >-Gregor

Ok. What I was planning on doing was letting the body rotate around the center of mass instead of the roll center. I have a hunch that the combination of jacking, lateral, longitudinal, anti-forces, and torques will happen to move the body right the way. After all, a large lateral force will lift the chassis more and the body won't end up rolling around the same roll center as before anyway since it'll move differently. I might drop you an email here before too long if you wouldn't mind discussing our methods further. I don't have the anti or jacking forces in yet, but I tried a really outrageous suspension system anyway once to see how it would work. The right rear wheel had double/triple length arms and the spring rates were all set to wierd values. The car ended up lifting the front right tire when cornering in one direction, then slamming down hard and lifting one of the rear tires when turning the other. In a straight line it was rolled perhaps 10 degrees and pitched quite a bit. Anyway, it looked really cool :-) Thanks, Todd Wasson --- Performance Simulations Drag Racing and Top Speed Prediction Software http://PerformanceSimulations.Com

GPL does not model caster.....I'm not sure what you're feeling there . N4 was Dave's first caster driving model. -- Doug Arnao www.vehiclecraft.com Ruud van Gaal <[email hidden]> wrote in message news:[email hidden]... > On Mon, 23 Jul 2001 17:56:00 GMT, "Stephen Smith" > <[email hidden]> wrote: > > >I'm just now getting around to mastering FF. Unlike every other FF-equipped > >racing sim, GPL models caster return. Thus, it's a remarkably effective > >teaching tool. If you're driving correctly (at least, if your steering > >input is correct), you'll never feel the caster return. If you're doing it > >wrong, you'll feel it: the c.r. practically rips the wheel out of your > >hands. > > > >[GPL's caster return also provides some otherwise-hidden input on how your > >shocks are set up. If you don't have the suspension properly damped, the > >caster return will oscillate, which you can definitely feel thru the wheel.] > > Ah, I always thought the oscillations were caused by the delay set too > high or low in the FF section. Now that I reread the previous > paragraph, no I never felt anything remotely like it! ;-) Uh... > > >Remarkable. We're still discovering new stuff in GPL 3 years on. > > It's amazing at the complexity that complex systems give; you can > never really understand the results your own product somehow, but it's > there. :) > > > Ruud van Gaal, GPL Rank +53.25 > Pencil art : http://www.marketgraph.nl/gallery/ > Free car sim : http://www.marketgraph.nl/gallery/racer/

>GPL does not model caster.....I'm not sure what you're feeling there . N4 >was Dave's first caster driving model. Was he using a predefined steer-camber "curve" based on caster or running the calcs around the actual kingpin axis, causing steer camber with both kingpin and camber angle? Thanks, Todd Wasson --- Performance Simulations Drag Racing and Top Speed Prediction Software http://PerformanceSimulations.Com

Hi Todd, "J. Todd Wasson" wrote:

Quoted text (9 lines)

> > Ok. What I was planning on doing was letting the body rotate around the > center of mass instead of the roll center. I have a hunch that the combination > of jacking, lateral, longitudinal, anti-forces, and torques will happen to move > the body right the way. After all, a large lateral force will lift the chassis > more and the body won't end up rolling around the same roll center as before > anyway since it'll move differently. I might drop you an email here before too > long if you wouldn't mind discussing our methods further. >

That's how you really should be doing things. The roll center description is really a steady state one, and the car should settle to this position if you apply the forces and torque correctly to the center of mass. So you hunch is perfectly correct. ;) What the roll axis does tell you is the direction of travel of the wheel relative to the body, which is perpendicular to this axis (vectorial product of the roll axis direction vector times any vector from the roll axis to the point on the wheel (is this the contact patch or the wheel center, I can't recall right now?)). The forces in the direction of wheel travel are transfered to springs, while those perpendicular to it are transfered directly to the body (via suspension links, but this doesn't interest you until you need to calculate suspension stress). Drop me an e-mail if you seriously believe I can answer some further questions :).

Quoted text (8 lines)

> I don't have the anti or jacking forces in yet, but I tried a really > outrageous suspension system anyway once to see how it would work. The right > rear wheel had double/triple length arms and the spring rates were all set to > wierd values. The car ended up lifting the front right tire when cornering in > one direction, then slamming down hard and lifting one of the rear tires when > turning the other. In a straight line it was rolled perhaps 10 degrees and > pitched quite a bit. Anyway, it looked really cool :-) >

Sounds a lot like the Renault 4 I got to drive once ;). -Gregor

That's not what he told me. You were there. ;o) --Steve "Doug Arnao" <[email hidden]> wrote in message news:KGo77.140866$[email hidden]...

Quoted text (16 lines)

> GPL does not model caster.....I'm not sure what you're feeling there . N4 > was Dave's first caster driving model. > > -- > Doug Arnao > www.vehiclecraft.com > > > Ruud van Gaal <[email hidden]> wrote in message > news:[email hidden]... > > On Mon, 23 Jul 2001 17:56:00 GMT, "Stephen Smith" > > <[email hidden]> wrote: > > > > >I'm just now getting around to mastering FF. Unlike every other > FF-equipped > > >racing sim, GPL models caster return. Thus, it's a remarkably

effective > > >teaching tool. If you're driving correctly (at least, if your steering > > >input is correct), you'll never feel the caster return. If you'redoing

Quoted text (7 lines)

> it > > >wrong, you'll feel it: the c.r. practically rips the wheel out of your > > >hands. > > > > > >[GPL's caster return also provides some otherwise-hidden input on how > your > > >shocks are set up. If you don't have the suspension properly damped,

the

Quoted text (19 lines)

> > >caster return will oscillate, which you can definitely feel thru the > wheel.] > > > > Ah, I always thought the oscillations were caused by the delay set too > > high or low in the FF section. Now that I reread the previous > > paragraph, no I never felt anything remotely like it! ;-) Uh... > > > > >Remarkable. We're still discovering new stuff in GPL 3 years on. > > > > It's amazing at the complexity that complex systems give; you can > > never really understand the results your own product somehow, but it's > > there. :) > > > > > > Ruud van Gaal, GPL Rank +53.25 > > Pencil art : http://www.marketgraph.nl/gallery/ > > Free car sim : http://www.marketgraph.nl/gallery/racer/ > >

Not sure Todd, but I believe he did modelled king pin inclination. -- Doug Arnao www.vehiclecraft.com J. Todd Wasson <[email hidden]> wrote in message news:[email hidden]... > >GPL does not model caster.....I'm not sure what you're feeling there . N4 > >was Dave's first caster driving model. > > Was he using a predefined steer-camber "curve" based on caster or running the > calcs around the actual kingpin axis, causing steer camber with both kingpin > and camber angle? > > Thanks, > Todd Wasson > --- > Performance Simulations > Drag Racing and Top Speed Prediction > Software > http://PerformanceSimulations.Com

Quoted text (19 lines)

>> Ok. What I was planning on doing was letting the body rotate around the >> center of mass instead of the roll center. I have a hunch that the >combination >> of jacking, lateral, longitudinal, anti-forces, and torques will happen to >move >> the body right the way. After all, a large lateral force will lift the >chassis >> more and the body won't end up rolling around the same roll center as >before >> anyway since it'll move differently. I might drop you an email here before >too >> long if you wouldn't mind discussing our methods further. >> > >That's how you really should be doing things. The roll center >description is really a steady state one, and the car should settle to >this position if you apply the forces and torque correctly to the center >of mass. So you hunch is perfectly correct. ;) >

Great! Thanks. Things will progress perfectly then :-) Look out world! Mu-ha-ha-ha-ha!! :-P

Quoted text (11 lines)

>What the roll axis does tell you is the direction of travel of the wheel >relative to the body, which is perpendicular to this axis (vectorial >product of the roll axis direction vector times any vector from the roll >axis to the point on the wheel (is this the contact patch or the wheel >center, I can't recall right now?)). The forces in the direction of >wheel travel are transfered to springs, while those perpendicular to it >are transfered directly to the body (via suspension links, but this >doesn't interest you until you need to calculate suspension stress). > >Drop me an e-mail if you seriously believe I can answer some further >questions :).

I think with my current system, this will all work itself out just fine. The only drawback is vertical tire flex can only be indirectly modelled by making a sequential spring system. Not the most accurate way, but this should be a big improvement over what I've currently got and it works really well with high-downforce setups. I'm thinking having accurate camber change, progressive spring travel, caster, bump steer, etc., would probably be better than having a linear system with a tire spring model that ignores all that or makes a rough guess at it. This first, then maybe a major rewrite later with everything. I might email you in time on a couple things. Thanks for confirming my hunches. Todd Wasson --- Performance Simulations Drag Racing and Top Speed Prediction Software http://PerformanceSimulations.Com

>Not sure Todd, but I believe he did modelled king pin inclination. > >-- >Doug Arnao >www.vehiclecraft.com > Thanks, Doug. Say, you wouldn't happen to have some geometric sketches and/or suspension data for one of your racing Porches laying around, would you? :-) I need a real suspension system to try out, so far I've got to make my own. If it's an easy mouse click away for you, I'd appreciate it. If not, no problem of course. Todd Wasson --- Performance Simulations Drag Racing and Top Speed Prediction Software http://PerformanceSimulations.Com