Code 93 Barcode Generator for Internal Identifiers

A tighter, self-checking step up from Code 39

Code 93 was designed as the denser, self-checking successor to Code 39, and it still sits between Code 39 and Code 128 on width. Each character occupies nine modules spread over three bars and three spaces, with a narrow gap between neighbours. That is fewer elements per character than Code 39 spends, so the same ticket number prints shorter while remaining readable by any scanner that has the symbology switched on.

The other half of the design is the checksum. Two check characters are added to every symbol: the C character, computed over the data with weights counting down from 20 to 2, then the K character, computed over the data plus C with weights counting down from 15 to 1. Both are taken modulo 47. This generator appends them without being asked and offers no switch to suppress them, which is deliberate: a truncated print or a dropped bar then fails the checksum instead of quietly handing your system a wrong number.

The trade-off is that Code 93 is not a GS1-registered symbology. There is no company prefix behind it and no catalogue that recognises it, so it belongs where one organisation controls both the label and the reader - stock rooms, library items, work-in-progress tickets, logistics sub-labels - rather than on products that scan at a till.

What the 47-character set lets you type

The Code 93 code table holds 47 values: ten digits, twenty-six uppercase letters, the space, and the six symbols - . $ / + %. This page accepts exactly those and nothing else. Lowercase letters, accented characters and punctuation such as commas, brackets or underscores produce an error rather than being silently remapped, so capitalise a value before you paste it.

The standard does reach further than 47 characters, using four shift symbols that encode an outside character as a pair; the table exists in the code we bundle, but the shift path is not exposed here. For mixed case or arbitrary ASCII, Code 128 is the simpler answer.

How the symbol is laid out

Every element in the symbol is a multiple of one narrow measurement, so the bar width control scales the whole code instead of only the bars. The symbol opens with a start pattern, carries your characters and the two check characters, then closes with a stop pattern and a final terminating bar. Because the check characters are derived from your data, they cannot disagree with it - only with a bad print.

One detail saves people confusion at sign-off: the line printed under the bars mirrors what you typed, while the bars themselves carry your data plus the two check characters. A scanner verifies those internally and normally strips them before the value reaches an application, so if the scanned output ends in two odd extra characters, the setting to change is on the reader.

Print and scan checks before the first run

The narrowest element in the symbol is the X-dimension, and everything else is built from multiples of it. Pick it from the room on the label and the printer you have, then change the bar width here rather than resizing a finished PNG, because scaling a bitmap softens the edges and eats into the clear space at both ends at the same time.

• Leave unprinted clear space at each end of the symbol; most internal specifications ask for around ten narrow-element widths, and the margin control adds plain background that counts towards it.
• Confirm the reader is enabled for Code 93 and for a data length that includes your value plus the two check characters.
• Print at the final size and scan that print - a screen preview flatters narrow bars.
• Keep the human-readable line so staff can key the value if the bars are damaged.
• If the identifier has to mean something outside your own organisation, use a format with registered meaning instead; Code 93 carries none.