The Teradata utilities expect both character and numeric data to be in "native" format on the client.
For numeric data, even the SMALLINT variable length indicators for VARCHAR fields or FASTLOAD-format records, the representation may differ depending on platform - and there is generally no easy way to adjust these between the export and the load. So you probably wind up CASTing each field individually to fixed CHAR in the mainframe FastExport, then using 'FORMAT UNFORMAT' on the non-mainframe load.
If your records contain exclusively fixed-length character data, there are two approaches: either use some external means to translate the text in between exporting and loading, or specify a compatible character set for the utilities to use on both platforms (e.g. UTF-8).