[ODE] approximation or rounding error?

[Bjoern Mahn]

hi,

I designed a small sphere fixed _in_ a big sphere. The small sphere is 
decentral, which should cause that the big sphere (lying on the ground) 
should move. Normally, it should have a behavior like a pendular, rolling 
from left to right and vice versa. But although the friction (.surface.mu) is 
dInfinity it oszillates higher and higher till the sphere rolles away.

Wherefrom comes the energy that causes this phenomenom? Are there rounding 
errors in the collision detection so that the rotation becomes more and more 
energy?

What did I do wrong?

(With using a quite big slip (FDS) this problem is fixed, but I don't want my 
sphere to be slipping so hard. Stepsize is 0.01)

Thanks for your help!

-- 
Björn Mahn