apparent increase in momentum for large time steps

[cekees]

I imagine this is user error, but I'm getting increases in momentum for a simple two body problem depending on the size of the time step even in Implicit Euler and HHT fails after repeated reduction of the time step. If I set the time step small enough with application->SetTimestep(); I can get it to go away, but I'm not sure how to set the time step a priori and was thinking chrono was doing something unconditionally stable and adapting dt based on error tolerances. Here is a video showing the issue:
chrono_momentum_conservation.avi.gz

Here is the code for solver config:

   // Change solver
    msystem.SetSolverType(ChSystem::SOLVER_MINRES);
    msystem.SetSolverWarmStarting(true);  // this helps a lot to speedup convergence in this class of problems
    msystem.SetMaxItersSolverSpeed(200);
    msystem.SetMaxItersSolverStab(200);
    msystem.SetTolForce(1e-10);

    // Change integrator:
    //msystem.SetIntegrationType(chrono::ChSystem::INT_EULER_IMPLICIT_LINEARIZED);  // default: fast, 1st order
    //msystem.SetIntegrationType(chrono::ChSystem::INT_HHT);  // precise, slower, might iterate each step
    msystem.SetIntegrationType(chrono::ChSystem::INT_EULER_IMPLICIT);  // precise, slower, might iterate each step