In some cases, r_rup is a negative number. This is a problem with the equation, not the implementation (confirmed it with Josh Steelman's excel example and it returns negative numbers for the same values). Josh Steelman is looking into this problem.

This is tough to estimate because Himmet has a more rigorous implementation of r_jb and r_rup, which would take 1 - 2 days to implement and debug. It would be easier to fix the current implementation.