Magnetostatic force on tiny mass objects - hitting machine precision

Hi
I have a study where I am looking at the magnetostatic force on a tiny object and the dimensions are such that I am hitting the machine precision in the force but the mass of the object is orders of magnitude smaller than the force and thus the resulting accelerations are essentially large and random due to the machine precision.

I have tried the finest mesh, the model is full 3D.

I will try adaptive mesh, putting in some symmetry but is it possible to adapt the force calculation to rescale to acceleration which will be well above the machine precision?

Regards
Jason