Unix Processing of IBM OS/390 Files

How does one convert non-printable characters [like nulls, hex ’00’] into Unix compatable data?We are migrating from an OS/390 to a Unix environment. A very important file contains non-printable characters but OS/390 Cobol reads the file. Unix sort apparently compresses out nulls and a succeeding Unix Cobol program will not read the file, losing track of the end of record. Help!!