smaller time steps diverge/incorrect

[jiajun]

Dear professor:
I am facing a problem when solving the stress equlibrium equation with density.  density*(partial^2(u)/partial(t)^2)=div(u) I use newmark and when I choose dt 0.005,the result is wrong and sometimes diverge. But when I choose dt 5,the result is right and the same as abaqus result. I am so curious .

[Prof. R.L. Taylor]

I assume you are writing your own element, since the program does not support what you are doing.
Can you describe your implementation and the example you are solving.  There is not enough information in your posting to understand what is wrong.