Car Physics: Mass or no Mass

As my simulation develops I am continually haunted by one single question; Should I give my wheels mass? At the moment I don't and the car body still has all of the subtle movements that one would expect .... What exactly am I missing apart from a huge headache trying to keep track of the trajectory of the wheel CM through it's complex arc, especially as I head off into the land of SLA with caster ... Having watched some of the replays in GPL it would seem that are keeping track of the wheels' mass, do they? TIA, Colin ----------------------------- http://www.cersoft.com

If you don't give the wheels mass they aren't going to respond to bumps and dips correctly. "Colin Reed" <[email hidden]> wrote in message news:[email hidden]...

Quoted text (20 lines)

> As my simulation develops I am continually haunted by one single > question; > > Should I give my wheels mass? > > At the moment I don't and the car body still has all of the subtle > movements that one would expect .... What exactly am I missing apart > from a huge headache trying to keep track of the trajectory of the > wheel CM through it's complex arc, especially as I head off into the > land of SLA with caster ... > > Having watched some of the replays in GPL it would seem that are > keeping track of the wheels' mass, do they? > > TIA, > > Colin > > ----------------------------- > http://www.cersoft.com

"Haqsau" <[email hidden]> wrote in message news:<a8am93$q7jof$[email hidden]>... > If you don't give the wheels mass they aren't going to respond to bumps > and dips correctly. You would think so yes but it really doesn't seem to make that much difference. Does GPL or N2002 model tire springs? Colin

"Haqsau" <[email hidden]> wrote in message news:a8am93$q7jof$[email hidden]...

Quoted text (9 lines)

> If you don't give the wheels mass they aren't going to respond to bumps > and dips correctly. > > "Colin Reed" <[email hidden]> wrote in message > news:[email hidden]... > > As my simulation develops I am continually haunted by one single > > question; > > > > Should I give my wheels mass?

Well yes...if the car in question is Italian. If it's British, then probably not...and if it's Hungarian...why are you building it? can cars be Catholic? dave henrie

"Colin Reed" <[email hidden]> wrote in message news:[email hidden]... > "Haqsau" <[email hidden]> wrote in messagenews:<a8am93$q7jof$[email hidden]>... > > If you don't give the wheels mass they aren't going to respond tobumps > > and dips correctly. > > You would think so yes but it really doesn't seem to make that muchdifference. Under normal circumstances perhaps not, but what about when cutting over kerbs? Seems like having mass would change how well the tire could follow the contour of the kerb, which would then change how much grip you would get from it. IOW a massless tire would allow somebody to cut over the kerbs with no real penalty to grip. Also remember wheel mass is part of your drivetrain inertia. > Does GPL or N2002 model tire springs? Don't know about those two, but I do know F1 2001 models tire spring rate, damper rate, and mass.

Hi Colin,

Quoted text (10 lines)

>As my simulation develops I am continually haunted by one single >question; > >Should I give my wheels mass? > >At the moment I don't and the car body still has all of the subtle >movements that one would expect .... What exactly am I missing apart >from a huge headache trying to keep track of the trajectory of the >wheel CM through it's complex arc, especially as I head off into the >land of SLA with caster ...

Hm, I have a setting in Racer which is close to this; debug.ini's use_susp_as_load. If on, it skips tire rate and uses the suspension instead (but still uses wheel mass to get the wheel's acceleration). I don't have a trajectory btw, just a pure vertical suspension system. For the moment I'll just have a simplistic form of trajectories, like camber change/mm suspension movement. The problem you might face is that with tire rate you need bigger frequencies (more physics updates/sec), which is why I originally did the use_susp_as_load setting, until the dampers were integrated implicitly instead of explicitly. I know use tire rates again, btw. It just explodes a bit sooner. >Having watched some of the replays in GPL it would seem that are >keeping track of the wheels' mass, do they? I don't know; that might have just been gyroscopic precession you're seeing. What GPL does have is fixed rollcenters, so you'd say no kinematic suspensions. So at least you can come quite far without trajectories. ;-) Perhaps Olley's calculations (the new book from Doug & Bill Milliken) can also help. Haven't bought it yet. Ruud van Gaal Free car sim: http://www.racer.nl/ Pencil art : http://www.marketgraph.nl/gallery/

> Under normal circumstances perhaps not, but what about when cutting over > kerbs? Seems like having mass would change how well the tire could > follow the contour of the kerb, which would then change how much grip > you would get from it. IOW a massless tire would allow somebody to cut > over the kerbs with no real penalty to grip. Ahhhh kerbs ;-) Can you imagine how cool it would be to be sliding sideways into a kerb, the two leading wheels hit and then boom, you are flipping in a beautiful barrel roll. > Also remember wheel mass is part of your drivetrain inertia. I have rotational mass because as you say they are part of the drivetrain inertia ;-) > Don't know about those two, but I do know F1 2001 models tire spring > rate, damper rate, and mass. Hmmm, I really should go and buy me a copy, I saw it in my local software shop recently ... hmmmm Colin

> Hm, I have a setting in Racer which is close to this; debug.ini's > use_susp_as_load. If on, it skips tire rate and uses the suspension > instead (but still uses wheel mass to get the wheel's acceleration). > I know use tire rates again, btw. It just explodes a bit sooner. Are you trying to imply that even using implicit integration, using tire rates introduces instabilities into the sim? > I don't have a trajectory btw, just a pure vertical suspension system. > For the moment I'll just have a simplistic form of trajectories, like > camber change/mm suspension movement. When I've finished I also won't have a trajectory as such because once the arm lengths and positions have been specified then the wheel will automatically move through the correct path of movement. > What GPL does have is fixed rollcenters, so you'd say no kinematic > suspensions. So at least you can come quite far without trajectories. > ;-) Indeed, but time to move on, I can't wait to get the effects of dynamic camber working but it should be a nightmare to setup, so many more parameters. Todd seems to be coming along really well, can't wait for a demo (is there one coming?) > Perhaps Olley's calculations (the new book from Doug & Bill Milliken) > can also help. Haven't bought it yet. Yeah tempting isn't it .... ;-) Colin

Quoted text (7 lines)

>Indeed, but time to move on, I can't wait to get the effects of >dynamic camber working but it should be a nightmare to setup, so many >more parameters. > >Todd seems to be coming along really well, can't wait for a demo (is >there one coming?) >

Thanks, Colin :0) No demo just yet. I don't use DirectX for anything 'cause this is written in PowerBasic, which doesn't have much in the way of DX wrappers. I just got Wav files to load in manually as a matter of fact, so I should be able to do a little sound mixing. GPL's Ferrari engine sounds GREAT at 15,500 rpm! lol I'll need to use standard Windows messages to allow controllers to be used. This works fine with my Thrustmaster SuperSport wheel with it's 15 pin connector, but I don't know if USB controllers will work this way.. Do you? I'll be using MM_JOY1MOVE, joySetCapture, etc... That won't allow force feedback, but will it allow most USB wheels to work anyway? What I really need is some sort of DirectInput wrapper Dll or something. I think I've got sounds pretty well licked, but it's done in software which is probably not as fast as DirectSound would be :-( Should get a wrapper for that too, or maybe switch to FMod like Ruud did. >When I've finished I also won't have a trajectory as such because once >the arm lengths and positions have been specified then the wheel will >automatically move through the correct path of movement I was under the impression you already had this running. Where are you right now with your suspension system? Todd Wasson --- Performance Simulations Drag Racing and Top Speed Prediction Software http://PerformanceSimulations.Com My little car sim screenshots: http://performancesimulations.com/scnshot4.htm

Quoted text (12 lines)

>> > If you don't give the wheels mass they aren't going to respond to >bumps >> > and dips correctly. >> >> You would think so yes but it really doesn't seem to make that much >difference. > >Under normal circumstances perhaps not, but what about when cutting over >kerbs? Seems like having mass would change how well the tire could >follow the contour of the kerb, which would then change how much grip >you would get from it. IOW a massless tire would allow somebody to cut >over the kerbs with no real penalty to grip.

You could still get this right with a massless tire, I think. The big effect there you're talking about is how the wheel might jump off the curb, causing a total loss of grip momentarily, right? This can still be done by noting the wheel mass in order to limit the suspension force on rebound, and at the same time ignoring it's effect in the regular loop of calculations (meaning essentially you're still using massless tires). You'd wind up with the vertical force going to zero if the wheel should leave the ground, but you wouldn't need all the complex and CPU hungry constraint stuff to do it, just a quick IF block to take care of it. Should get the best of both worlds this way, 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

> GPL's Ferrari engine sounds GREAT at 15,500 rpm! lol Yeah, the first time I got the sound working I deliberately disconnected all drive to the wheels and just played with revving the engine for about an hour, sad isn't it but you should have seen the smile on my face ;-) > I'll need to use standard Windows messages to allow controllers to be used. > This works fine with my Thrustmaster SuperSport wheel with it's 15 pin > connector, but I don't know if USB controllers will work this way.. Do you? > I'll be using MM_JOY1MOVE, joySetCapture, etc... That won't allow force > feedback, but will it allow most USB wheels to work anyway? I just ripped loads of code from one the DX SDK samples and shimmied it into my engine, there is still a bug in it where either dinput or dsound is causing the application to not terminate properly !?!?!? > I was under the impression you already had this running. Where are you right > now with your suspension system? Ahhh alas no, currently I have massless wheels (but with rotational inertia) coupled with a vertical ray casting "lets find where the ground is and if it is less than the spring rest length then put the wheel there" kind of thing ... I then find the vertical velocity of the wheel by doing vertVel = (currlen - oldlen) / dTime. A total hack but I wouldn't be suprised if most games so far use this method ... I am currently toying with a new method which I call "valid configuration" suspension which allows me to specify limits in the suspension travel as well and should be a whole load less hacking than now. Colin

>> GPL's Ferrari engine sounds GREAT at 15,500 rpm! lol > >Yeah, the first time I got the sound working I deliberately >disconnected all drive to the wheels and just played with revving the >engine for about an hour, sad isn't it but you should have seen the >smile on my face ;-) LOL! I did the same thing with my other system, I know how it goes :0) Life's little triumphs and pleasures... >I just ripped loads of code from one the DX SDK samples and shimmied >it into my engine, there is still a bug in it where either dinput or >dsound is causing the application to not terminate properly !?!?!? Beats me... Maybe Ruud, Ashley, or Gregor would know. (Matt Jessick would too I'm sure, but he's been scarce lately.) Anyway, it sounds like your suspension works pretty much the same way mine does. I still need to convert it to truly use the full 3-D geometry so anti-dive/squat, etc., are included. Right now it's just a front view virtual arm system, but it works great and is almost as fast as using a vertical wheel travel model. You're right though, it makes car setup much more complicated. I counted about 16 different ways at each wheel to adjust camber, for instance, and they all have slightly different impact on the handling.. Sheesh.. lol Todd Wasson --- Performance Simulations Drag Racing and Top Speed Prediction Software http://PerformanceSimulations.Com My little car sim screenshots: http://performancesimulations.com/scnshot4.htm

> Anyway, it sounds like your suspension works pretty much the same way mine > does. :-) > I counted about 16 different ways at each wheel to adjust camber, for instance, > and they all have slightly different impact on the handling.. Sheesh.. lol Do you have cast setup yet? Colin

>Do you have cast setup yet? Caster? Yep, all setup, but it doesn't change with wheel travel yet. For that, I need the full 3-D system. Are you doing McPherson struts or A-arms in yours first? Todd Wasson --- Performance Simulations Drag Racing and Top Speed Prediction Software http://PerformanceSimulations.Com My little car sim screenshots: http://performancesimulations.com/scnshot4.htm

> Caster? Yep, all setup, but it doesn't change with wheel travel yet. > For that, I need the full 3-D system. Doh!!! Yes I did mean caster ;-) Stupid keyboard ... stupid hands .... > Are you doing McPherson struts or A-arms in yours first? A-arms first, they are easier, I want to get the suspension module setup first then I can just add suspension types as I think of them. Colin

You need to decide what dynamics you will be simulating. If you include the wheel mass as a separate object supported by a tire spring then the major new thing you get is the "Wheel Hop" mode. This is resonance at 10 to 15 cycles per second of the car and wheels system. Controlling this mode to keep the tires planted as much as possible is one of the jobs of the shock absorbers. If you don't include the mode, you don't need to do "fast bump" etc. modelling of the shocks for example, because the car object (sprung mass) won't usually move (or oscillate) that fast. This mode is much higher than the frequency of the sprung mass on the suspension springs, and so can give stability or modelling trouble when integrating. Colin Reed wrote:

Quoted text (8 lines)

> > > Hm, I have a setting in Racer which is close to this; debug.ini's > > use_susp_as_load. If on, it skips tire rate and uses the suspension > > instead (but still uses wheel mass to get the wheel's acceleration). > > I know use tire rates again, btw. It just explodes a bit sooner. > > Are you trying to imply that even using implicit integration, using > tire rates introduces instabilities into the sim?

Using implicit integration, if you are integrating at too low a frequency then the dynamic mode just gets filtered out. Kind of a "tree falls in the forest" type of thing: If it goes into the integrator and doesn't come out can you really say you are "modeling" it? ;) - Matt