This issue came up a couple of years ago, and we had some similar problems.
The Berkeley group suggested putting a plug of silver wool (10-30 mm) in
the bottom of the combustion column. In addition, we have found that
replacing the "water" trap every 400 samples or so also is required. With
BOTH of these two modifications, the HCl seems to be trapped, removed, and
nitrogen isotopes run OK, i.e. reproducible and precise.
These are good, fast and cheap things to do.
Marilyn Fogel