Matrix Sans Fonts

On this page:

Intro โ–ช Prior designs โ–ช Beyond ASCII โ–ช My past designs โ–ช Honourable mentions โ–ช Results

Research

Much of my interest in the topic was piqued by Damien Guardโ€™s article, Typography in 8 bits: system fonts. In preparation for this project, I decided to conduct an even more comprehensive study of classic dot-matrix fonts, with a focus on those using the iconic 5×7 dot matrix capitals. In the end, over 40 existing 5×7 font designs were studied.

These example fonts are presented in approximate chronological order, ranging from the 1960s to the present day. This list is by no means exhaustive; it simply contains all the examples I could find.

Many of the fonts depicted below contained a full set of 128 or 256 characters, including special glyphs for pseudo-graphics applications, other scripts, etc. So as to compare the glyph designs themselves rather than the character sets, only the standard ASCII ranges of each font are reproduced here (equivalent to Unicode's Basic Latin block). The original systems employed character cells of various sizes, resulting in different spacing between characters and lines. For clarity and consistency, the fonts have been presented here with 7×10 dot character cells.

I have arranged my comments on the fonts in PMI format: Plus (pros), Minus (cons) and Interesting. The earlier examples tend to attract more comments not necessarily because they are more idiosyncratic, but because each particular quirk is mentioned here only once. Much more could be written about each font if they were critiqued individually.

All images on this page are made available under the Creative Commons Zero licence (public domain dedication)

Prior designs

IBM 029 Card Punch

IBM 029 character set

Reference: Norbert Landsteiner — Punched Card Typography Explained

Font replicating this design: Keypunch029 by Stewart C. Russell

The IBM 029, a machine for creating punched cards, was introduced in 1964 to replace the 026 Printing Card Punch from 1949. I decided to show the 029 character set here because it could print a few more characters than the 026. They are the two oldest dot-matrix character sets I have found in my research. They are so old, in fact, that the patterns were stored on a metal “code plate” since electronic read-only memories — indeed, integrated circuits (or silicon chips) in general — were yet to be invented.

Despite their advanced age, a majority of these glyphs have stood the test of time, appearing the same as newer 5×7 dot matrix fonts. There are only a couple of truly unusual designs, like the letter S which was probably chosen to differentiate it from the number 5.

P ๐Ÿ˜ M ๐Ÿ˜• I ๐Ÿค”
  • Clear, open dollar sign $ with small s to allow the strokes to protrude above and below
  • Heavy punctuation marks ' , . : ; are distinguishable in adverse conditions
  • Clear, exaggerated curves of brackets/parentheses ( )
  • Serifs differentiate number 1 and letter I
  • High-waisted 5 is well-differentiated
  • Squarish G has good legibility
  • Wavy diagonals in per cent sign % and X
  • Heavy apostrophe ' is ugly
  • Crowded dots in asterisk *
  • Slightly low hyphen/minus sign -
  • Semicolon ; not aligned with colon :
  • Narrow question mark ?
  • Serifs on B D help to differentiate from 8 O, but inconsistent with the other letters
  • Short crossbars on E F
  • Upper terminal of G does not match C
  • Serifless J does not match I
  • High diagonal of N
  • Diagonal centre of S
  • Stumpy central peak of W
  • Unusual, decorative number sign #
  • Dots in %; no room for circles
  • Floating diagonal in slash / looks odd today, but a nearly universal choice in 5×7 fonts
  • Narrow zero 0 to differentiate from O
  • Exaggerated lower curve on 2
  • Somewhat awkward diagonal on 7
  • Large less- and greater-than signs < >
  • At sign @ visibly based on a, but loop does not extend all the way around
  • Angular A
  • Somewhat awkward shape of M
  • Rectangular O to differentiate from zero 0
  • Sharp V is effective but does not match A

Project Intrex

Project Intrex display console character set

Reference: Paul Allen King Jr. — A novel solid state character generator

While searching Internet Archive for information, I found a 1969 thesis about the development of a character generator for an automated library system at MIT. This device supplied character patterns for a computerised video display, still quite a novel concept at the time. The dot matrix data were stored in a hand-woven core rope memory.

The character set depicted here was actually not intended to be used directly, but instead with interpolation circuitry that smoothed curves and diagonals to produce a 9×13 dot matrix. As such, many of the designs are unconventional or suboptimal in appearance in their 5×7 form.

P ๐Ÿ˜ M ๐Ÿ˜• I ๐Ÿค”
  • Full-height stroke in dollar sign $
  • Effective six-pointed asterisk *
  • Semicolon ; and colon : properly aligned
  • 5×5 less- and greater-than signs < > (albeit with wavy diagonals)
  • Exaggerated curve of D differentiates from O
  • Serifs on J harmonise well with I
  • Conventional design of S
  • Effective, if oversized, grave accent `
  • Attractive f uses full width, but low crossbar
  • Highly exaggerated curly brackets/braces { } very effective and well-differentiated
  • Asymmetrical slash in per cent sign %
  • Apostrophe ' resembles prime mark; would not work as opening quotation mark
  • Full-height slashes / \ leading to lumpy diagonals
  • Poorly-executed slashed zero 0
  • Elongated central terminal of 3
  • Top-heavy 5
  • Tails of 6 and 9 too long
  • Poorly-executed question mark ?
  • Uneven at sign @
  • Pointy A v
  • Uneven weight in diagonals of M N
  • Reduced x-height of b d
  • Bowl of g does not match b d p q
  • Dot of i higher than that of j
  • Serifless l (small L) hard to differentiate
  • Rounded m inconsistent with n
  • Imbalanced y
  • Small number sign #
  • Large, uneven upper part of ampersand &
  • Straight lower terminal of 5
  • Straight diagonal of 7
  • Aggressively-rounded C only matches D
  • Prominent crossbar of G
  • Very sharp V leads to narrow crotch
  • Excellent angular shape of W w but risks looking like a shaded pattern, losing legibility
  • Low caret ^
  • Asymmetrical c
  • Floating g p q with single-dot descenders
  • Unusual but effective design of k
  • Tails of n q r u
  • Gap in vertical bar |

Texas Instruments TMS4103

Texas Instruments TMS 4100 character set

Reference: Centronics 101 brochure

The TMS4103 was an early MOS character generator: that is, a special ROM chip containing character bitmaps. These devices were standard parts stocked by their manufacturers, without the need for a custom order as with ROMs in general. The TMS4103 dates back to at least 1970, when it was used in the first dot-matrix printers, the Centronics 101 and DEC LA30 DECwriter (well, I assume it was also used in the 101, since its character set is identical to that of the LA30).

P ๐Ÿ˜ M ๐Ÿ˜• I ๐Ÿค”
  • Consistent slashes in % / \ (but still ugly)
  • Large, clear comma ,
  • Single-dot punctuation marks . : ; look better in running text
  • Slashed zero 0
  • Colon : on baseline
  • More usual question mark ? shape
  • & has flair but slightly uneven weight
  • Awkward 2
  • Very awkward 3
  • Narrow semicolon ; spacing
  • Oddly rounded circumflex ^
  • Wide gap in exclamation mark !
  • Prominent uptick and serifs on 1
  • Extra-large less- and greater-than signs < >
  • At sign @ mostly encircled, but no a visible
  • Round-topped A
  • Horizontal terminals on S
  • W like upside-down M is highly legible
  • Rounded Y with centred junction

Fairchild 3258

Fairchild 3258 character set

Reference: 3258 datasheet

Another character generator from the early 1970s.

P ๐Ÿ˜ M ๐Ÿ˜• I ๐Ÿค”
  • Sharp and legible 2
  • Centred diagonal of N looks a bit odd, but is the best design yet in my opinion
  • Flat-topped 3 aids differentiation from 8, but looks to be missing dots in the centre
  • Wide colon : spacing
  • Mismatched colon : and semicolon ;
  • Tight upper curve of question mark ?
  • Similar at sign @ fully encircled
  • Straight-sided V (a common choice)
  • Round-bottomed W (also popular)
  • Lower crotch of Y, shape echoes V

General Instrument RO-5-2240S

General Instrument RO-5-2240S character set

Reference: RO-5-2240S datasheet

Yet another character generator. The glyph patterns are similar to those of the Fairchild 3258 above, with only a few differences.

P ๐Ÿ˜ M ๐Ÿ˜• I ๐Ÿค”
  • Effective design of tilde ~ (in place of backslash \)
  • More subtle curly brackets { }
  • Aggressively-rounded at sign @; circle touches “a”

Mostek MK2408P

Mostek MK2408P character set

Reference: MK2408P datasheet

Still another character generator.

P ๐Ÿ˜ M ๐Ÿ˜• I ๐Ÿค”
  • Attractively curved 2, a popular style
  • Small less- and greater-than signs < > uncommon but very effective
  • Round G consistent with other letters
  • Comma-like design of apostrophe ' would not work as an opening quotation mark
  • Two-dot commas , ; possibly too subtle
  • Upper diagonal of 3 better, but still ugly
  • Top-heavy 7
  • Very abstract at sign @
  • Diagonals of V look like tapered curves
  • Interesting (and probably unique) ampersand & design
  • Oversized asterisk * (but a common choice)
  • D matches other curved glyphs, but differs from O by only two dots
  • Low crossbar on G carries typographic flair, but looks out of place to my eye

National Semiconductor MM4240

National Semiconductor MM4240 character set

Reference: MM4240 application notes

This is a composite of two character sets: those of the MM4240AA (capitals only) and MM4240AE (small letters only) character generators. Like most such devices from the era, the glyphs are strictly confined to the 5×7 dot matrix, forcing the descending letters g j p q y to sit high on the baseline. The designs of the small letters appear to have been influenced by traditional handwriting; note especially a b d e g l p q y.

Apart from the slash / and 7, all the characters up to Z are identical to those of the IBM 029. There seems to have been a trend of including a crosshatch graphic in the 0x7F position, also seen in several of the character sets below. This could serve as a blinking cursor on video displays.

P ๐Ÿ˜ M ๐Ÿ˜• I ๐Ÿค”
  • Attractive bowl shape of b d g p q
  • Consistent curves of b d g h n p q r u y
  • Sophisticated, well-executed k
  • Well-executed t is balanced, uses full width
  • v matches V
  • Centres of m w match
  • Awkward diagonal on 7
  • Bowls of d g upside-down (d needs a spur, g doesn’t; see example of convention)
  • Crossbar of f too short, does not match t
  • Ugly tail of i and hook of j
  • Rounded corner of q
  • Narrow curly brackets { } hard to recognise
  • Awkward tilde ~
  • Script-style a with tail
  • Open c is highly legible but looks wide
  • Rounded e with short tail
  • Mirrored i j
  • Handwritten-style l
  • Short centre leg of m

Motorola MCM6674

Motorola MCM6674 character set

Reference: MCM6674 datasheet

Font replicating this design: Another Man’s Treasure by Rebecca G. Bettencourt

This character generator is notable for having being used in the Tandy/Radio Shack TRS-80 Model I computer. It was relatively advanced in that it contained glyphs for all 128 possible ASCII codes, including lower-case letters.

Apart from O and S, the capitals are identical to those of the IBM 029.

P ๐Ÿ˜ M ๐Ÿ˜• I ๐Ÿค”
  • Consistent curves of b d g h n p q r u
  • Ugly comma design ' , ;
  • Inconsistent vertical alignment between colon and semicolon : ;
  • Odd design of grave accent `
  • Narrow f does not match t
  • Ugly curves of f t
  • j looks too wide; lacks a serif to match i
  • v w both inconsistent with V W
  • Curve of y inconsistent with b d g h n p q r u
  • Tilde ~ less recognisable and raised up, which limits its usefulness
  • Large, heavy, eight-pointed asterisk *
  • Fully-rounded bowls of small b d g p q
  • Narrow, single-dot aperture in c
  • Curled tails of g y
  • Narrow k

Signetics 2513

Signetics 2513 character set

References: Signetics 2513 and GI RO-3-2513 datasheets

This is another composite of two character sets, belonging to the 2513/CM2140 (upper case) and 2513/CM3021 (lower case) character generators. The 2513 was a common part that was also produced by other manufacturers like General Instrument. This chip, with the upper-case-only character set, was used by the Apple I and Apple II computers.

P ๐Ÿ˜ M ๐Ÿ˜• I ๐Ÿค”
  • Narrow, straight apostrophe ' matches double quotation mark "
  • Effective script a
  • Another abstract at sign @
  • Wide and bold square brackets [ ]
  • Poorly constructed k
  • Very low central terminal of w
  • Very narrow f t
  • Flat curves of h n r u y

National Semiconductor DM8678

National Semiconductor DM8678 character set

Reference: DM8678 datasheet

This is also a composite of character sets, from the DM8678CAB (upper case) and DM8678CAH (lower case) character generators. The DM8678CAH still contained only 5×7 dot matrices for each glyph, but shifted the descending letters g j p q y by two lines during output in order to display them correctly. This has the side-effect of making j appear sunken, with its dot at the waistline of the font.

P ๐Ÿ˜ M ๐Ÿ˜• I ๐Ÿค”
  • Clean, consistent lower case alphabet, with the sole exception of j
  • Excellent question mark ?
  • Consistent, legible b c d g h n p q y
  • Another awkward, top-heavy 7
  • Serifs on upper terminals of G T
  • Hook of j matches neither g y nor f t
  • Long central leg of M
  • Short tails of g y
  • Narrow r (looks more proportional to me; compare with those above and below)
  • Shorter t matches conventional typefaces better, but less legible
  • Spurless u, a la Futura

Mullard SAA5050

Mullard SAA5050 character set

Reference: SAA5050 datasheet

Font replicating this design (with totally smooth diagonals): Bedstead by Ben Harris

This chip was designed for Teletext applications and also saw use in home computers such as the BBC Micro. A full nine dots of height are available here, allowing for descenders that properly match the ascenders. The SAA5050 was able to interpolate these glyphs, to produce a 10×18 matrix with less jagged diagonals.

P ๐Ÿ˜ M ๐Ÿ˜• I ๐Ÿค”
  • Comma and semicolon , ; extend below baseline, taking advantage of space
  • Colon and semicolon : ; properly aligned for lower-case text
  • Odd dollar sign $
  • Narrow hyphen - to differentiate from dash in “English” character set version
  • Unusual pointed zero 0 differentiates from capital O without using a slash, which may have confused general audiences at the time

Trendcom 200

Apple Silentype character set

Reference: Silentype manual

The Trendcom 200 was a thermal printer designed for use with personal computers. It is best known for being rebranded as the Apple Silentype, Apple’s first printer. It has one of the more interesting character sets on this list.

P ๐Ÿ˜ M ๐Ÿ˜• I ๐Ÿค”
  • Another more reasonably sized asterisk *
  • Unique and effective bold comma in , ;
  • Rare and effective tail of R
  • Non-descending g j p q y fit between baseline and waistline
  • Short terminals of s
  • Curves of 6 9 match 0 O etc.
  • Low crossbar of A
  • Unique diagonals of M W
  • Narrow c looks more proportional (but has inconsistent spacing with adjacent letters)
  • W w are both rotated forms of M m, leading to unique m shape and central terminal of w being longer than usual

Apple IIe

Apple II character set

Reference: KreativeKorp — Ultimate Apple II Font

Font replicating this design: Print Char 21 by Rebecca G. Bettencourt

The first 64 glyphs (the first two rows here) are lifted from the Signetics 2513/CM2140 character generator, used in the earliest Apple computers.

The latter 32 glyphs were added to the Apple III (1980) and subsequently the Apple IIe in 1983 (with the crosshatch at 0x7F replacing the Apple logo from the Apple III font). The larger character cells on these systems allowed for descenders on the small letters, if somewhat stunted at only one dot below the baseline.

P ๐Ÿ˜ M ๐Ÿ˜• I ๐Ÿค”
  • Another clean, consistent lower-case alphabet, with the exception of g u
  • k 5 dots wide; very open and legible
  • g too round
  • m w too square
  • r u do not harmonise with h n y
  • Unusual, effective length of central leg of m
  • Bold but evenly-curved tilde ~; a good solution to keep it two dots high

Motorola MC6847

Motorola MC6847 character set

Reference: Wikipedia

Font replicating this design: Hot CoCo by Rebecca G. Bettencourt

This video chip, with built-in character generator, was used in the Tandy/Radio Shack TRS-80 Color Computer. It contains a set of negative (or inverse, i.e. white on black) capitals, which there was no need to reproduce here, instead of small letters. Notice that this set is nearly identical to that of the IBM 029 Card Punch seen above.

P ๐Ÿ˜ M ๐Ÿ˜• I ๐Ÿค”
  • Clumsy-looking 4 6 9
  • Diamond in place of asterisk *

Texas Instruments 99/4A

TI 99/4A character set

Reference: Classic99 emulator (see screenshot)

The TI 99/4A was another early home computer. Its character set has the rare inclusion of small capitals instead of lower-case letters; this may have been a compromise decision as the character cells are not deep enough for descenders. The small caps are generally well-executed and perfectly match the full capitals.

P ๐Ÿ˜ M ๐Ÿ˜• I ๐Ÿค”
  • Effective bold comma and semicolon , ;
  • Regular S in dollar sign $, with another non-protruding stroke
  • Narrow j k
  • Non-slashed, full-width zero 0

Oric

Oric character set

Reference: Proposal to add characters from legacy computers and teletext to the UCS

The character set of the Oric series of home computers is almost identical to that of the Apple IIe, except for the altered comma, 2, 6, 9, colon, semicolon and square brackets, plus a few special characters in the non-alphanumeric positions.

P ๐Ÿ˜ M ๐Ÿ˜• I ๐Ÿค”
  • Aggressively-rounded corners of copyright symbol ยฉ mean the circle touches the c
  • Long, low stroke on pound sign ยฃ

EACA Colour Genie

Colour Genie character set

Reference: Colour Genie firmware (see graphic)

Another, less well-known, home computer from the early 1980s.

P ๐Ÿ˜ M ๐Ÿ˜• I ๐Ÿค”
  • Crossbar of f on waistline, as in traditional typefaces
  • Very long serif of J
  • Serifs of T Z inconsistent within the font
  • Elongated t
  • Unique ampersand &
  • Serifless 1
  • Descender of f
  • Effective tilde ~ (but 6 dots wide is cheating)

COMX-35

COMX-35 character set

Reference: COMX-35 firmware (see graphic)

This obscure home computer has one of the worst character sets I’ve seen (the ampersand & really does sit below the baseline; that wasn’t a mistake on my part!). There would have been plenty of room for lower-case letters if the set didn't have duplicate glyphs in different colours, multiple variations of the COMX logo and other useless pseudo-graphic characters. What’s more, the RCA CDP1870 video chip supports character cells 9 dots high, which would have allowed for lower-case with true descenders.

P ๐Ÿ˜ M ๐Ÿ˜• I ๐Ÿค”
  • Imbalanced number sign #
  • Drooping upper terminal of 2
  • Extra dots in 5 A
  • Short terminals of 9 G

Aquarius

Aquarius character set

Reference: Aquarius firmware (see graphic)

The Aquarius computer, manufactured by Radofin, is best known for being briefly sold under the Mattel brand. The first 64 glyphs in this set are identical to those in the upper-case Signetics 2513/CM2140 character generator (except the apostrophe!) and the latter 32 appear to have been derived from the lower-case 2513/CM3021 with some modifications.

P ๐Ÿ˜ M ๐Ÿ˜• I ๐Ÿค”
  • Uneven arms on k
  • m w 6 dots wide and uneven
  • Short descenders of p q
  • Square y
  • Cruciform t

Commodore MPS-803

Commodore MPS-803 character set

Reference: printout from my MPS-803

The MPS-803 was a dot-matrix printer designed for use with Commodore home computers. Being an 8-bit Commodore device, its built-in character set does not adhere to the ASCII standard, so this image has been rearranged for presentation here.

This has one of the cleanest, most consistent capital alphabets yet examined, but the small alphabet suffers from numerous quirks.

P ๐Ÿ˜ M ๐Ÿ˜• I ๐Ÿค”
  • Round A G make for consistent look
  • Unusual and effective y based on v shape, as in conventional typefaces
  • Narrow p q
  • Uneven s
  • Tail on a

BASIC Level III (Sega SC-3000)

BASIC Level III (Sega SC-3000) character set

Reference: SC-3000 Survivors online emulator (see screenshot)

The SC-3000 was a computer version of Sega’s first game console, the SG-1000. Like the SG-1000 but rarely for home computers, it has no built-in ROM for firmware, instead relying on plug-in cartridges. This character set belongs to the BASIC Level III cartridge.

P ๐Ÿ˜ M ๐Ÿ˜• I ๐Ÿค”
  • Well-executed yen sign ¥
  • Effective Greek letter pi π
  • Serif and weak diagonal on 7
  • Drooping upper terminal on question mark ?
  • Square a
  • Wide crossbars on f t
  • Chunky 4 is highly legible
  • Rounded a within at sign @
  • Short central terminal of W

MSX

MSX character set

Reference: La Nueva Escuela — Making an MSX font

Font replicating this design: MSX Screen 0 by Andy Teijelo

Inspired by successful industry standards like VHS and disappointed in the Wild-West fractured home computer market, several (mostly Japanese) manufacturers adopted a common 8-bit computer standard, MSX. The full character set is similar to that of the IBM PC, with the glyphs redesigned to fit the TMS9918 video chip’s text mode.

P ๐Ÿ˜ M ๐Ÿ˜• I ๐Ÿค”
  • Poorly-executed ampersand &
  • Low slashes / \
  • Angular 5
  • Serif and odd diagonal on 7
  • Centred stem of f leaves little room for hook
  • Crowded dots in g p q
  • Extra dot in middle of s
  • Bold less- and greater-than signs < >
  • D combines both legibility enhancements
  • Tail on u

Elektronika BK-0010

Elektronika BK0010 character set

Reference: BK Back to Life emulator (see screenshot)

This is a rare example of a home computer from the Soviet Union. Overall, it has one of the nicest fonts on this list, with relatively few foibles. It also contains glyphs for the Cyrillic (Russian) alphabet, of course, so it can be a useful example for extending my font later.

P ๐Ÿ˜ M ๐Ÿ˜• I ๐Ÿค”
  • 7-dot high version of the angular y
  • Asymmetric quotation marks " '
  • Unique and ugly at sign @
  • Large currency sign ¤
  • Rarely, B D do not match
  • Extra-pointy curly brackets { }

Samsung SPC-1000

SPC-1000 character set

Reference: Wikipedia

The SPC-1000 was Samsung’s first personal computer. It used the AMI S68047 video chip, similar to the Motorola 6847 and evidently with a very similar built-in character generator ROM. The latter 32 glyphs here are unique to the SPC-1000, however. Note that the relative letter spacing has been altered in the graphic above to accommodate a few 7-dot wide glyphs; this system used 8×12 character cells.

P ๐Ÿ˜ M ๐Ÿ˜• I ๐Ÿค”
  • Long upper terminal and ugly tail of a
  • Won sign โ‚ฉ uses width to advantage

Motorola MC6847T1

Motorola MC6847T1 character set

Reference: Wikipedia

This updated video chip was used in the late-model Tandy/Radio Shack TRS-80 Color Computer 2. It now contains a lower-case alphabet, with true descenders. Some of the unusual glyph designs have been cleaned up in this version, although some have also been made worse.

P ๐Ÿ˜ M ๐Ÿ˜• I ๐Ÿค”
  • Another asterisk * with uneven weight
  • Missing dot in centre of 3
  • At sign @ unique and ugly
  • j y shorter than other descending letters
  • Humpless m

Sega barcodes

Sega barcode numerals

References: Altered Beast and Black Belt case inserts

These numerals were found alongside the barcodes on some Sega Master System games. They contain some interesting features that I haven’t seen anywhere else. Some of those may have been inspired by the typeface OCR-B, which is normally used on EAN and UPC barcodes; that would explain the designs of the digits 1, 3, 4, 6 and 9 (but not 2, 5 or 7).

P ๐Ÿ˜ M ๐Ÿ˜• I ๐Ÿค”
  • Effective, serifless 1
  • Well-executed flat-topped 3 (finally!)
  • Open counter of 4 is attractive and legible
  • Odd diagonal of 2
  • Short tails of 6 9 are legible but appear somewhat imbalanced

Hitachi HD44780U

Hitachi HD44780U character set

Reference: HD44780U Datasheet

This LCD controller, dating back to the 1980s, has been widely used in character LCDs. Particularly popular in hobbyist applications and commercial devices of low-volume production, there are hundreds of these modules available from dozens of manufacturers. Nowadays, such devices usually emulate the HD44780U’s behaviour and character ROM. Of the character LCDs available on Mouser at the time of writing (that list the glyph patterns in the datasheet, like this example), all appear to use nearly identical glyph designs, usually varying by only a few dots here and there. They are likely derivatives of each other, with differences probably due to errors in manual transcription. Shown here is the A00 ROM, which provides the option of true descenders (and very deep ones too, at 3 dots below the baseline).

P ๐Ÿ˜ M ๐Ÿ˜• I ๐Ÿค”
  • Cluttered yen sign ¥
  • Uneven c
  • Very deep descenders of g j p q y

Terminal (Microsoft Windows)

app850.fon character set

Reference: app850.fon (system font file)

This font, aptly named Terminal, is used when a Command Prompt window is set to use the 6×8 raster font (those being the full character cell dimensions, including the gaps around the 5×7 glyphs). It dates back at least as far as 1992, in Windows 3.1’s DOSAPP.FON. It has quite a number of unusual features compared to the other fonts in this list. The design strikes me as something that was adapted from larger dimensions (the file does contain fonts of various sizes, possibly lending some credence to this idea), which may explain the unconventional design, odd proportions and inconsistent weight of certain glyphs.

P ๐Ÿ˜ M ๐Ÿ˜• I ๐Ÿค”
  • Wide spacing in equals sign =
  • Narrow Z h k n u x y z ~
  • Only one serif at bottom right of i l but regular serif on j
  • Serifs on r
  • Uneven weight of w y
  • Bold exclamation mark !
  • Short number sign #
  • Diagonal stroke in dollar sign $
  • Narrow round brackets ( )
  • Small, bold asterisk *
  • Another style of bold comma ,
  • High central terminal in W
  • No diagonals in x
  • Extra dot in centre of curly brackets { }

FF Dot Matrix Two

FF Dot Matrix character set

Reference: MyFonts — FF Dot Matrix

The earliest example I have seen of a commercial typeface imitating the 5×7 dot matrix style, dating from the early 1990s. Around that time, foundry FontFont also popularised another utilitarian style with their famous release, FF DIN.

P ๐Ÿ˜ M ๐Ÿ˜• I ๐Ÿค”
  • Excellent slash and backslash / \ use the full height without appearing lumpy or wavy
  • Another ugly ampersand &
  • Mismatched comma and full stop , .
  • Asymmetrical tilde on ~
  • Small, cross-shaped asterisk *
  • At sign @ uses broken ring to good effect
  • Descender of Q

Fairfax

Fairfax character set

Reference: KreativeKorp — Fairfax

This open-source font by Rebecca G. Bettencourt was created for the stated purposes of coding and plain text editing, and has a wide Unicode coverage with support for some constructed scripts in the Unicode Private Use Area. Similarly to the font I wish to create, it is distributed as an outline font in TrueType format, having been converted from bitmaps using custom software.

Fairfax has the best lower-case b d g p q that I have seen in any 5×7 font. It is the only design examined here that has the conventional bowl shapes for these letters (see the National Semiconductor MM4240 section above). While still being restricted to 5 dots of width, some glyphs enjoy a much taller dot matrix, such as the quotation marks, slashes and brackets. This affords the font a more modern feel, unlike anything derived from one of the “classical” 5×7 fonts above.

P ๐Ÿ˜ M ๐Ÿ˜• I ๐Ÿค”
  • Large dollar sign $ with protruding stroke
  • Tall brackets, slashes, vertical line () [] {} / \ |
  • Gentle curve of 7
  • Well-executed 5×5 less- and greater-than signs < > like rotated v
  • Narrow curly brackets { } much more effective with this height
  • Per cent sign % contains real circles instead of dots, but this harms its clarity
  • Quotation marks " ' raised above cap line
  • Small, raised number sign #
  • Asterisk * raised but with usual shape
  • Flat upper serif of 1
  • Unusual 6 9
  • Variation of the abstract at sign @ manages to contain recognisable a
  • Asymmetric c

PixelMix

PixelMix character set

Reference: pixelmix.ttf via DaFont

This is a popular 5×7 dot matrix font by Andrew Tyler, freely available online for personal use, with over 400,000 downloads on DaFont at the time of writing. The design appears to be based on the A02 ROM of the Hitachi HD44780U, with most of the alphanumeric characters matching it. While the small letters are therefore a bit wonky and inconsistent, it does have the very effective slash and backslash seen in FF Dot Matrix above.

P ๐Ÿ˜ M ๐Ÿ˜• I ๐Ÿค”
  • Ideal size and shape of grave accent `
  • Wide and ugly ampersand &
  • Plus-shaped asterisk *
  • Slightly narrow, lop-sided question mark ?
  • Extra wide at sign @
  • Extra tall f
  • g rises above waistline
  • Wide stroke spacing in number sign #
  • Right-angled comma , ;
  • Square corners of curly brackets { }

5by7

5by7 character set

Reference: 5by7.ttf via peter-wiegel.de

This open-source font by Peter Wiegel was inspired by his first font project: modifying the character set of a laboratory instrument to support German umlauts. Although the glyphs are strictly limited to a 5×7 matrix, it is a proportional font, so narrow glyphs have only one dot of spacing.

P ๐Ÿ˜ M ๐Ÿ˜• I ๐Ÿค”
  • Another unusual ampersand &
  • Asterisk * is a square
  • Bold exclamation mark !
  • Different shape of non-descending y

Minecraft

Minecraft character set

Reference: ascii.png (game texture file)

Although this is by far the most stylised font shown here, I felt it was worth examining since it belongs to the best-selling video game of all time; countless words in this font have been read by hundreds of millions of people around the world. This makes it a significant design with wide recognition, especially amongst young people.

In monospaced fonts like those above (except PixelMix), it was normal to include serifs on 1 I i l for extra width. The Minecraft font is proportionally spaced, however, so many glyphs are much wider or narrower than usual. Additionally, the proportions of many glyphs (like the small punctuation marks, small raised asterisk, full-height slashes, etc.) are unusual for 5×7 dot matrix fonts, but perfectly normal for conventional typefaces. These features all help to give the Minecraft font a more contemporary feel.

P ๐Ÿ˜ M ๐Ÿ˜• I ๐Ÿค”
  • Yet another poorly-executed ampersand &
  • Small, raised asterisk * less recognisable
  • Wide base of 1
  • Serif on 2
  • Tailless 4
  • Long tails of g y
  • Square corner of w
  • Narrow dots in percent sign %
  • Larger at sign @ still an abstract squiggle
  • High-waisted capitals A B E F G H K P R S X Y
  • Stroke of Q only extends outwards
  • Vertical tail of R
  • Inconsistent terminals of S
  • Serifless i j l

Beyond ASCII

There has long been more to computer fonts than the 95 printable ASCII characters. Some of the fonts examined above contained accented letters and other special characters, which we will now explore further. Where necessary, I have rearranged the glyphs to match the order of ISO‑8859‑1 (and thereby Unicode's Latin‑1 Supplement block) to make comparison easier.

See how the diacritical marks are very large, compared to conventional typefaces, owing to the number of dots required to define them.

Mullard SAA5050

Mullard SAA5050 extended character set

This set is a composite of the various language versions of the SAA5050 that were offered.

P ๐Ÿ˜ M ๐Ÿ˜• I ๐Ÿค”
  • Effective pound sign ยฃ
  • Attractive section sign ยง is clearly based on S
  • Wide degree sign ยฐ keeps counter open, even after application of the SAA5050’s diagonal-smoothing algorithm
  • No bar in fractions ยฝ ยผ ยพ to maximise legibility at this low resolution
  • Attractive sharp s รŸ
  • Distinguished, right-angled cedilla on รง
  • Diacritics of ร  รข รจ รฉ รช รด touching base glyphs
  • Square currency sign ยค
  • Shorter letters ร„ ร… ร‰ ร– รœ รฌ รฎ รฒ รป รผ to fit diacritical marks
  • Single dot for ring of ร… รฅ

Level III BASIC (Sega SC-3000)

BASIC Level III (Sega SC-3000) extended character set
P ๐Ÿ˜ M ๐Ÿ˜• I ๐Ÿค”
  • Ugly pound sign ยฃ
  • No right downstroke on micro sign/mu ยต
  • Inverted question mark ยฟ does not match regular one
  • Acute and grave accents on ร€ ร are droopy
  • Misaligned diacritics on รˆ ร‰ รŠ ร‹ รŒ ร รŽ ร’ ร“ ร›
  • Inverted exclamation mark ¡ on baseline, suits all-capitals text
  • Diagonal cedilla on ร‡

MSX

MSX extended character set
P ๐Ÿ˜ M ๐Ÿ˜• I ๐Ÿค”
  • Broken stroke of cent sign ยข differs from $
  • Another unsuccessful attempt at a calligraphic pound sign ยฃ
  • Ordinal indicators ยช ยบ have different widths
  • Narrow counter of micro sign/mu ยต
  • Fractions ยผ ยฝ ยพ inconsistent and cramped
  • Width of รผ inconsistent with รน รบ รป
  • Prominent crossbars of yen sign ยฅ
  • Large central section of narrow section sign ยง
  • Curved cedilla on รง
  • ร• รต differ only in width

Hitachi HD44780U

Hitachi HD44780U extended character set

These glyphs are from the A02 ROM version of the HD44780U, which lacks descenders but does contain most of the ISO‑8859‑1 character set.

P ๐Ÿ˜ M ๐Ÿ˜• I ๐Ÿค”
  • Well-executed, calligraphic pound sign ยฃ
  • Micro sign µ matches u, as in most conventional sans-serif typefaces
  • Awkward section sign ยง
  • Ugly pilcrow/paragraph sign
  • Serifed thorn Þ þ
  • Tall æ
  • ì í î ï do not match i
  • Ugly (but at least clear) eth ð
  • Very large degree sign °

Command Prompt (Microsoft Windows)

app850.fon extended character set

These glyphs come from the system file app850.fon, (presumably named for Code Page 850) as it contains all the characters of ISO‑8859‑1.

P ๐Ÿ˜ M ๐Ÿ˜• I ๐Ÿค”
  • Effective copyright symbol ©
  • Small superscript ³ is barely legible
  • Horizontal grave and acute accents À Á etc. are only recognisable in context
  • Tilde Ñ etc. looks more like double acute accent
  • Ugly Ø ø
  • Greek delta δ in place of eth ð
  • Square ring of Å å
  • Rotationally symmetric æ

Fairfax

Fairfax extended character set

This font employs full-height accented capitals, with separate (not conjoined) diacritical marks. This greatly enhances its consistency, legibility and modernity, not to mention its suitability for writing languages other than English.

P ๐Ÿ˜ M ๐Ÿ˜• I ๐Ÿค”
  • Perfectly consistent and clear letters (with the only exceptions of the awkward Ø ð)
  • Very angular micro sign/mu µ
  • Extra dot near centre of pilcrow/paragraph sign
  • Unique, well-executed pound sign £
  • Single stroke in yen sign ¥
  • Unusual but effective section sign §
  • Incomplete circles in copyright and registered trademark symbols © ® to remain 5-dots wide

Minecraft

Minecraft extended character set

This font makes use of wider glyphs as well as taller ones, giving it the largest character cells shown on this page.

P ๐Ÿ˜ M ๐Ÿ˜• I ๐Ÿค”
  • Well-executed eth ð

My past designs

I created my first dot-matrix font in 2013, using pencil and graph paper. I then scanned and digitised it with the help of Microsoft Paint. Ever since then, tinkering with such designs has been an occasional pastime for me. Initially, I did this with relatively little outside inspiration, but over time my work lost some originality in favour of more conventional (and usually better) design choices. Most were not based on 5×7 dot matrices, but below are four selected designs that were.

There is little to say about these designs that hasn’t already been mentioned above, but I will note some of the more interesting features:

Untitled, 2014

My past design No. 1

This was one of my earliest efforts, based on the 7×7 font I created first. The long ascenders and descenders, number 1, capital letters A I S V Y and small j are all unusual designs. The small letters almost have the traditionally correct shapes, if not for d.

Untitled, 2014

My past design No. 2

This is an improved version of the previous design, now based on smaller character cells. Here, I had arrived upon my favourite design of the slashes / \, which admittedly doesn’t work very well in the per cent sign %. There is also a unique design for the capital G and the rare variant of R, which I must have created independently. I remember adding the extra curves to S after seeing other examples and realising my previous effort was rather unusual.

Untitled, 2018

My past design No. 3

This design was created from scratch, without reference to any previous efforts. All the alphanumeric characters here are strictly 5 dots wide, including 1 I i j l. We can see the introduction of my favourite design for the question mark ?, and the capital Y no longer resembles V.

QS Matrix 1.0, 2019

My past design No. 4

This design was created primarily as a font for the Quikscript and Shavian alphabets, hence the name QS Matrix. The design may have been based on the previous one, as it shares many similarities. The small y is unusual, and was inspired by the one from the Commodore MPS-803 printer as seen above.

Version 2 of this font was made after beginning the study documented here. As a result, I decided to alter some glyphs to make the font appear a little more generic.

Honourable mentions

I only discovered the following examples after this study was complete and all key glyphs for Matrix Sans were already designed. Nevertheless, in the interests of completeness, here are all of the remaining 5×7 dot matrix fonts that I have found (excluding most modern designs that can be found on websites such as FontStruct, as they are high in quantity and low in quality). Most of them still have many unique and unusual features:

Texas Instruments TMS9918 data book

Texas Instruments TMS9918 sample character set

Reference: TMS9918 Data Book

From the data book pertaining to the video chip used in countless computers and video game systems of the 1980s. The small capital-style lower case is similar to that seen in the font from the TI 99/4A computer (see above).

Sinclair QL

Sinclair QL character set

Reference: Sinclair QL "PM" ROM (see graphic)

Mid-1980s business-oriented microcomputer.

Memotech MTX512

Memotech MTX512 character set

Reference: MEMU emulator (see screenshot)

Mid-1980s home computer. The lower-case alphabet contains a number of oddities.

Monaco

9-point Monaco bitmap font

Reference: Infinite Mac emulator

Designed by Susan Kare, this is the 9-point version of the monospaced bitmap font provided with early Apple Macintosh computers. It contains a number of very unusual features (particularly # % 6 9 Q R X) and was probably a fully original design, rather than being largely derivative like almost every other font shown on this page.

SAM Coupรฉ

Sam Coupรฉ character set

Reference: SimCoupe emulator (see screenshot)

Late-1980s home computer. The font is very similar to that of the Sinclair QL (see above), with most of the differences being in the lower case letters.

Neckerspoel

Neckerspoel font

Reference: necker.mf from “capbas” font collectiion

A Metafont font dating from the early 1990s. According to the author, the name Neckerspoel โ€œcomes from the futuristic bus station in Eindhoven, where they use it in the โ€˜flip-dotโ€™ displays above the buses.” The characters $ 3 ? ? @ W a i l w are unique or unusual.

Dottie

Dottie font

Reference: Dottie font from MyFonts

An early-2000s commercial typeface.

Results

I attempted strike a balance between four principles when designing my own 5×7 font:

The former two qualities are often in conflict with the latter two; for example, the crossbar of small f should ideally be at the waistline, but almost never is in the fonts seen above. Here, I would make the crossbar lower in the interests of authenticity. Doing so also adds a degree of quirkiness, which I think is desirable in such a font.

Of course, there is really a fifth, hidden criterion for choosing these designs: personal taste. Despite attempts to remain impartial, some of my preferences — even dating back to my earliest, inexperienced designs — have sneaked into this project. Most notable are the rare less-than and greater-than signs < > and question mark ?, and the (possibly) unique small m. To be fair, these instances can actually be justified under the sound design principle, as sometimes the most common variant seen above just isn’t very good (in my opinion). But perhaps it is only right that I do leave some kind of mark here, rather than creating the most generic design possible.

So, after seeing a dizzying array of 5×7 dot matrix fonts from past and present, it is finally time to design my new font. I consider the following glyphs to be easy choices, because the overwhelming majority of existing fonts use them or because they have no reasonable alternatives:

Glyphs with easy design choices

While this image only spans the range of the ISO‑8859‑1 character set, I plan for the font to support both Adobe Latin 4 and Google Fonts Latin Expert Google Fonts Latin Level 4 Google Fonts Latin Core-Vietnamese-Plus, which have much wider Unicode coverage. Also note that I have had to extend the character cells to 7 dots wide in order to fit the vulgar fractions ยผ ยฝ ยพ, and the copyright and registered trademark symbols © ®, which require 6 dots of width. It is possible to make those fractions only 5 dots wide, as in the Mullard SAA5050 seen above, but that scheme would not work for the others like โ…” โ…œ โ… โ…ž.

To fill in the gaps of the above figure, I judged which option would best suit my intended aesthetic for the font, based on the four principles mentioned at the start of this section and visual trials of different variations. I consider many of these options to be equally good. In fact, in some cases my real personal preference ended up as the second choice, since I ultimately decided a different option was more suitable for the font as a whole. The following table lists the glyphs I chose from, in decreasing order of preference (i.e. with my final decision leftmost):

Options Rationale
Option 2 is too big, so much so that it might confuse contemporary audiences.
Tests proved the bold punctuation marks don’t suit the font as a whole. This choice also applies to others like the comma, colon, etc.
The 3-dot hyphen suits texts better and the 5-dot variant makes a better minus sign. Since this Unicode character does double duty, the 4-dot variant is a compromise.
I am fond of Option 2, but it is so rare that the authenticity principle dictated Option 1 in this case.
A more understated design and far more common.
Chosen as it appears less sharp and angular, better fitting with the other digits.
Option 2 is more legible and better distinguished from 8 and the Cyrillic letter ะ—, but probably too rare to use here.
The smaller less-than and greater-than signs are a better fit with both the other mathematical operators and conventional typefaces.
Option 3 lacks the point of inflection that I think question marks should have and Option 2 has a cramped upper curve, failing the sound design principle.
I like Option 2, but Option 1 is the more common design and looks better in the context of email addresses, etc.
Option 2 here (as with D G below) is more consistent with the other capitals and arguably a more sound design. I chose Option 1 to inject a little more quirkiness to the font.
Reasons as noted for A above. Option 1 is also much more legible.
Reasons as noted for A above. Options 3 and 4 are perhaps a little too exaggerated for this font.
Chosen for consistency with capital I.
Both have imperfect diagonals. I’ve changed my mind on this one over the years, but at the moment Option 2 looks stranger to me.
Less legible than Option 2, but chosen for consistency with A and v.
Also less legible than Option 2, but chosen for consistency with V v w
The more common design, which I also feel looks slightly nicer.
A tough decision. Option 3 matches option 2 of h n r u, which I don’t like very much. Option 2 is the most well-designed; perhaps too well-designed for this font. Therefore, I chose Option 1 for being suitably quirky, yet functional.
Although Options 2 and 3 are more legible, I prefer glyphs that are 5 dots wide and Option 2 is unattractive.
The shorter terminal makes the letter look clearer to me.
Is 5 dots wide, more open and legible, more authentic, and the hook is consistent with j t.
I much prefer the look of option 1. It is also consistent with b d etc.
The hook is consistent with f t and the serif is consistent with i.
Is 5 dots wide and more open and legible.
This feels most natural to me and is a compromise between the other, equally common, options.
Option 2 would match Option 2 of b h etc.
Similar reasons as for f.
Option 2 looks like the two sides are curved, rather than diagonal. Option 1 has straight sides but clear diagonals near the bottom.
Chosen for consistency with m.
Another tough decision. I like option 2 and its effect on the typographic colour of text in the font, but it is rather rare and other people don’t seem to like it as much.
Option 1 is highly exaggerated, but I find the others too hard to recognise as curly brackets.
A more understated design that is more consistent with the rest of the font.
Same reasons as for the pound sign £ above.
Chosen for consistency with the pilcrow/paragraph sign ¶.
Option 3 has no separation between the pair and Option 2 is too stylistic.
A more traditional Greek shape, and unrelated to the design of u as it technically should be.
The longer version has better visual proportions.
I feel that Option 1 best suits the overall character of the font.

With those more difficult choices made, here is the final design I arrived upon. In my view, it contains all of the best features from the many designs examined above and none of the worst:

Final design of Matrix Sans

This may be the most effort anyone has ever put into a classic dot matrix-style font. I hope you find the result useful!