EBCDIC: Encyclopedia II - EBCDIC - Technical details

EBCDIC - Technical details

EBCDIC code pages and ASCII-based code pages are incompatible with each other. Since computers only understand numbers, these codepages assign a character to these numbers. The same byte values are interpreted as a different characters depending on the codepage used. Data stored in EBCDIC require a code page conversion before the text can be viewed on ASCII based machines, like a personal computer.

A single EBCDIC byte occupies eight bits, which are divided in two halves or nibbles. The first four bits is called the zone and represent the category of the character, whereas the last four bits is called the digit and identify the specific character.

There is a nice correspondence between hexadecimal character codes and punch card codes for EBCDIC. This was an important feature at the time the EBCDIC scheme was created. An IBM card punch could make a 12-row punch card with up to 2 punches per column, the first punch somewhere in the first 3 rows (called the zone) and the second punch somewhere in the last 9 rows (called the number). The zone could thus be considered a value from 0 to 3, and the number a value from 0 to 9, where 0 means no punch, and non-zero means the corresponding row was punched. The initial version of EBCDIC was just (0xf-zone)<<4+number and defined only the lower-left 10x4 part of the table shown below (the zone was apparently reversed so the letters would at least be in alphabetic order).

There are a number of different versions of EBCDIC, customized for different countries. Some East Asian countries use a double byte extension of EBCDIC to allow display of Chinese, Japanese and Korean scripts for their mainframes. In the double byte extension of EBCDIC, there are shift codes [0x0E,0x0F] to shift between the single byte and double byte modes.

IBM typically names all of its code pages with a number called a CCSID (Coded Character Set IDentifier). It is important to note that the same CCSID can have different character positions in a codepage. For example, the newline character can be a different byte value in z/OS UNIX System Services versus the other EBCDIC based operating systems. This becomes an issue when transferring EBCDIC based text data between machines.

Adapted from the Wikipedia article "Technical details", under the G.N U Free Docmentation License. Please also see http://en.wikipedia.org/wiki