Following my previous post, I've just released DSSR beta-r03-on-20130309. I tested this version against all nucleic-acid-containing structures in the PDB, so all the bugs you reported yesterday should have been fixed. I was actually approaching the end of such an extensive tests when I say your post, and your reported cases were already handled at that time.
It's worthy noting that some of the DSSR buggy cases are due to problematic data quality of the corresponding PDB entries. For example,
1bdn apparently have two structures overlapped (with 4 chains instead of 2) -- check the structure with PyMOL or Jmol, and you will seem my point.
While more bugs are bound to be uncovered, and further refinements are necessary, I am confident to say that DSSR (beta-r03-on-20130309) is robust enough for real-world applications. As always, do not be shy to report back any issues you encounter.
Xiang-Jun