[ODE] QNAN in Body position! May be LCP problem? Need Help!!!

[DjArcas]

Bah, me again, still investigating, as this problem has been causing me
serious problems.

If I run at a timestep of 1/60, I get a QNAN and Zero length vector
immediately.

If I run at a timestep of 1/120, it's fine.

If I run two timesteps of 1/120, one immediately after the other, I get the
same problem.

To say I'm perplexed is an understatement. What could make running two
timesteps one after the other a problem, but running it once be ok?

Adam

----- Original Message ----- 
From: "Andrey Tyagunoff" <moonrat@nikita.ru>
To: <ode@q12.org>
Sent: Saturday, June 21, 2003 5:02 PM
Subject: RE: [ODE] QNAN in Body position! May be LCP problem? Need Help!!!


> Situation changed!
>  I have test ONLY 2 bodyes with 2 binded geomsSphere.
>
> dWorld settings:
> dReal = double CFM = 1e-10 ERP = 0.2 gravity = (0, 0, -9.8)
>
> Bodyes have mass initial position gravitation etc.
> First case:
> Bodyes linked by joint Hinge2. -> have a QNAN in positions  on 2 sec of
> game cycle.
> Second case:
> Bodyes NOT linked by joint Hinge2. -> have a normal  game cycle.
>
> How I linking the bodyes:
>   m_hJoint = dJointCreateHinge2(getWorld(), 0);
>   dJointAttach(m_hJoint, hBody1, hBody2);
>
>   const dReal* const wPos = dBodyGetPosition(hBody1);
>   dJointSetHinge2Anchor(m_hJoint, wPos[0], wPos[1], wPos[2]);
>   dJointSetHinge2Axis1(m_hJoint, 0, 0, 1);
>   dJointSetHinge2Axis2(m_hJoint, 1, 0, 0);
>
>   dJointSetHinge2Param(m_hJoint, dParamLoStop, 0);
>   dJointSetHinge2Param(m_hJoint, dParamHiStop, 0);
>   dJointSetHinge2Param(m_hJoint, dParamFMax, 50);
>   dJointSetHinge2Param(m_hJoint, dParamVel, 0);
>   dJointSetHinge2Param(m_hJoint, dParamVel2, 0);
>   dJointSetHinge2Param(m_hJoint, dParamFMax2, 80);
>
>   dJointSetHinge2Param(m_hJoint, dParamSuspensionERP, 0.25);
>   dJointSetHinge2Param(m_hJoint, dParamSuspensionCFM, 0.005);
>
>
> And WHAT'S WRONG????
>
> -----Original Message-----
> From: ode-admin@q12.org [mailto:ode-admin@q12.org] On Behalf Of Andrey
> Tyagunoff
> Sent: Saturday, June 21, 2003 4:30 PM
> To: ode@q12.org
> Subject: RE: [ODE] QNAN in Body position! May be LCP problem? Need
> Help!!!
>
>
> All wheels in car have mass. And after star car succesful landing to
> heighmap. But wheels of car small shaking on land.
>
> -----Original Message-----
> From: DjArcas [mailto:djarcas@hotmail.com]
> Sent: Saturday, June 21, 2003 4:27 PM
> To: Andrey Tyagunoff; ode@q12.org
> Subject: Re: [ODE] QNAN in Body position! May be LCP problem? Need
> Help!!!
>
>
> QNANS can be caused by not giving an object mass, or by giving it a mass
> of
> zero.
>
> ----- Original Message ----- 
> From: "Andrey Tyagunoff" <moonrat@nikita.ru>
> To: <ode@q12.org>
> Sent: Saturday, June 21, 2003 1:04 PM
> Subject: [ODE] QNAN in Body position! May be LCP problem? Need Help!!!
>
>
> > I create four wheels car with Hinge2 joints.
> > And I have heightmap custom geom for land.
> > After start the aplication all looks good, but some times my car
> > disappear. In this case dGetBodyPosition() return me QNANs!
> >
> > If i set #define DEBUG_LCP I have message
> > "L*D*L' check, maximum difference = 4.7683716e-007"
> > If I trying use double in my config settings I have this message too.
> > What's wrong???
> >
> >
> >
> > _______________________________________________
> > ODE mailing list
> > ODE@q12.org
> > http://q12.org/mailman/listinfo/ode
> >
>
> _______________________________________________
> ODE mailing list
> ODE@q12.org
> http://q12.org/mailman/listinfo/ode
>
> _______________________________________________
> ODE mailing list
> ODE@q12.org
> http://q12.org/mailman/listinfo/ode
>