PK violation capture

hi all,

I am trying to load a table from some flat files.

Is it possible to redirect only those rows that violate the PK constraint to a badData file and keep the good records going into my table?

I tried implementing a Data Flow Task that has a FF source, a Data Conversion step and a SQL Server (OLE DB) Destination with a flat file on the error output of the data conversion block, however, the process as seen in the Output, it loads all records, then catches the PK violation and cleans up everything - hence mnothing in my table, nothing in the badData file ...

thanks,
kowalsky