The Early Years
aving visited the excellent Teletext Then and Now site created by Darren Meldrum and Mike Brown, I remembered that I had some off-screen shots of early teletext reception, so I dug them out and made up this page of reminiscences around them. Later I came across a pile of long-forgotten indexes, press releases and other items that I'd received from BBC and IBA Engineering Information in the seventies. I parcelled them all up and sent them off to Mike, who has incorporated some of them into his site.
Long distance television reception was a hobby of mine from the sixties, and when teletext arrived on the continent I was keen to try to receive it. My first success was a Nederland 2 test transmission.
I also found some BBC Microcomputer programmes that I'd copied down from CEEFAX pages into a notebook, so I've included those, together with the photographs, here on my own web site.
If you're interested in any aspect of teletext, you may wish to subscribe to the UK Teletext Mailing List.
our tv picture consists of a single spot of light created by an electron beam which sweeps quickly from left to right to form lines, and more slowly from top to bottom to form frames. In the first half of the century, when the tv systems were designed, it took receivers a relatively long time to force the spot to "fly back" from the right-hand side of the screen to the left, and from the bottom to the top. Gaps were left in the transmitted signal that contained only black space and "blacker-than-black" synchronising pulses.
Modern sets have much shorter flyback periods, and more efficient blanking circuits, which means that the transmitted blanking period no longer needs to be black. In the sixties broadcasters started to use the "spare" lines for test signals, and then began considering what else they might put there.
In the early seventies, the BBC was experimenting with optional subtitles, and the IBA with source identification (SLICE), both of which have found their way into the present teletext services. They then realised the service would be more useful if it carried more general news and information for viewers to read. After some experiments the BBC (CEEFAX, first broadcast as a thirty-page magazine on 23 September 1974) and IBA (ORACLE) in consultation with BREMA, the set manufacturers' organisation, jointly produced the 1974 Broadcast Teletext Specification. In 1976 a second specification was published allowing for more display attributes and a modified character set, among other things.
The key to teletext is that the receiver contains a memory, to store data transmitted over a period of several frames, and a character generator to present the data on the screen. That means that data for a whole row of text (40 characters) can be transmitted in one tv line, and data for a whole page spread over several frames. At first only two lines per field (four lines per frame) were allowed, and a 24-row page took about a quarter of a second to send. Nowadays twelve lines per field is the norm, and many more pages are transmitted.
n 1975/76 Wireless World published a design for a teletext decoder that could be built by a home constructor with little or no test equipment. The design used about a hundred TTL (transistor-transistor logic) ICs on three printed circuit boards, and a kit was made available. The intention was to tap off a video signal from a colour tv, and feed the decoded RGB text signal back to the set. Interfacing was quite tricky, especially as sets of that period did not have the smooth video response they do today. Later, other magazines published designs that took an aerial signal and modulated the teletext to be fed into the aerial socket of a tv set, like a vcr does. The displayed text quality was not as good as with an integral decoder.
I built the WW decoder in 1976 and fitted it into a 1973 Pye 697 hybrid chassis. In those days of valves, it was not usual to apply RGB to the cathodes of the display tube. Instead a powerful video output valve fed luminance (Y) to the cathode, and smaller valves fed the colour difference signals (R-Y, G-Y, B-Y) to the grids. Because the colour difference signals were lower bandwidth than the luminance this saved on wideband power output stages. It also saved "matrixing" the colour signals to produce RGB - a process that's all done on a chip these days.
Because the matrixing was done "in the tube" it meant there was no RGB path in which to insert the teletext, so it was necessary to apply an inverted signal to each colour difference circuit. Unfortunately, the restricted bandwidth led to fuzzy characters, and a circuit was included to widen the white parts of the letters so they became brighter and more distinct. In 1978 I constructed a Forgestone colour tv, which, while still being hybrid in the sense that it had valves in the line output stage, had solid state RGB drivers, and so was easier to interface, and gave sharper results.
aving constructed the Wireless World Teletext Decoder, and incorporated it into the Pye colour set, I took some off-screen photographs. A couple of years later when I moved the decoder to the Forgestone set I took some more. The quality is not brilliant, but the shots give an idea of what was being broadcast in 1976-78, so here they are, together with some memories that they have jogged. There's also a page of photographs of some Dutch teletext received here in Sheffield in November 1979.
eletext and the BBC Computer Literacy Project were closely linked in 1982 (the year the Project and the associated BBC Microcomuter were launched). The intention was for CEEFAX to broadcast pages of computer-readable data for downloading via a teletext decoder attached to the BBC Micro. (Why do we not call them "micros" any more? Presumably it's no longer "PC"!) At first, the programmes were short, and transmitted as ASCII listings, and so it was possible to copy them down in longhand. Just as well, as there was no decoder available at the time, and there was a six-month waiting list for the BBC Micro itself, even for those (like me) who sent off their £400+ the day it was launched in February 1982.
I jotted down five such programmes in my "How to learn BBC BASIC" notebook, and they have survived both my BBC Micro and its associated discs and tapes. I have typed them into my Archimedes A440/1, and they work, so I present them here for your perusal, together with screen shots.
HE BEAUTY of teletext is that text is transmitted as character codes which is economical with bandwidth and allows the decoder to deal with the rendering of the actual glyphs (character shapes) at whatever resolution the designer feels is appropriate. However, because character generator chips are mass-produced, only a few styles of character have been used during the thirty years of decoder production. Almost all use the dot-matrix method of rendering text, switching sharply between the foreground and background colours as determined by the character generator ROM. Some advanced decoders use a system called 'anti-aliasing' whereby some of the pixels are a blend of the foreground and background colours which results in slightly blurred, but smoother, less jagged glyphs.
There are several basic ways of rendering text on an interlaced display like a television set. The easiest is to use a 5 x 9 matrix and repeat the pattern on adjacent fields so that pairs of lines in the resulting 5 x 18 display are identical. A refinement of this is to interpolate the 5 x 9 matrix and add half-sized 'fillets' to smooth the diagonals. This results in very slight flicker because alternate fields are different. Another way is to remove interlace altogether and superimpose the lines of alternate fields so that a 5 x 9 flicker-free display is produced. Only the full-screen display can be treated that way. Subtitles and mixed screens are rendered over the normal interlaced raster. Many of the decoders using this technique have higher-resolution 10 x 9 glyphs. Finally, it is possible for the decoder to rasterise fonts using anti-aliasing software which results in different patterns on alternate lines, giving higher definition glyphs that are subject to a degree of flicker, though this is minimised by the use of shades in between the foreground and background colours. Digital tv receivers and of course modern computers use this method almost exclusively.
This is the basic character glyph found on the early decoders such as the Wireless World design that used standard computer character generator ROMs.
The actual space allocated to each character is 6 x 10 pixels (6 x 20 for both fields) and so the decoder inserts a row and column of blank pixels to keep the characters apart.
Here the same character generator is used but a character rounding circuit has been added to improve the diagonals.
This was how text in the later versions of the WW decoder and Mode 7 of the BBC Microcomputer looked.
In more modern decoders character generators with twice the horizontal resolution have been used, though with still only nine elements vertically.
The same character generator operated in non-interlaced mode, with alternate fields superimposed instead of interleaved. In fact the character generator stores data for all 12 x 10 pixels so that large symbols and diacritical marks can be accommodated more easily, albeit with the danger of characters 'bleeding' into each other. Character code 96 (_), the 'wide bar' or 'm-dash', is such a culprit, though a row of them may be used to make a dividing line so: _________________________ which on an older decoder would appear dashed, so: --------------------.
It is noticeable that Ceefax now uses this method of constructing horizontal lines in favour of using block graphics characters, presumably so that pages may be converted automatically to 'digital text' format, where block graphics are not available.
This is the sort of output you would expect to see from a character generator with anti-aliasing.
Not only is the horizontal resolution improved (there are about sixteen columns at teletext character size) but text looks smoother because shades in between the foreground and background colours are used.
In addition, the data on alternate lines is different, yielding eighteen rows of resolution at this text size at the expense of some flicker on thin horizontals.
People find this sort of text easier to read and quote it as the main reason why 'digital text' is better than FLOF teletext, though it would be quite possible to implement such a display in a standard teletext decoder.
Teletext has a basic graphics facility using a 2 x 3 matrix of blocks that is the same size as a text character.
On the left is a solid ('contiguous') character, and on the right the same character in separated ('non-contiguous' or 'mosaic') mode.
DVB Text and subtitling
UMPING AHEAD to the twenty-first century, it may be worth digressing for a moment to see how text and subtitling are carried on the new digital television services.
Analogue satellite transmissions, which came along in the 1980s, can carry the same teletext magazines as an ordinary terrestrial service. The data rides on the spare lines just above the top of the picture as usual.
Digital transmissions (DVB, for 'Digital Video Broadcasting') can also carry exactly the same teletext pages, including subtitles, but they are encoded separately and sent mixed in with the audio, video and 'housekeeping' data streams. A digital decoder can do one or both of two things: it can convert the data to standard teletext format using the spare lines for your telly to decode and display as before, or it can decode the data itself and display it on the screen - useful for recording subtitles etc.
But digital transmissions may contain a new kind of text and/or subtitling service. Teletext was designed to be extremely simple to decode and sparing of bandwidth - consumer digital electronics was still in its infancy in the 1970s and there is a limited amount of free space in the analogue television signal. The solution adopted was that each character on the screen took one data byte to transmit, and its position in the data stream corresponded directly with its position on the screen.
By contrast, digital text uses a so-called 'mark-up' language - just like pages on the World-wide Web - and the decoder generates pretty lettering according to the verbose instructions it receives. Consequently, more colours can be used, text is smoother, and graphics (including photographs etc) can be displayed. However, a lot more data has to be sent per page than with teletext.
Unfortunately, the text systems at present used are not universal - in the UK alone, a different system is used for terrestrial and satellite - and a third system is being introduced on the continent of Europe. However, the DVB subtitling system is more standardised. In it, the shape and colour of the lettering is defined by the broadcaster and so subtitles on the various channels can appear very (or just subtly) different from each other. In general, however, they are all generated 'on the fly' by the broadcaster from the same original data as the teletext subtitles.
Unfortunately, because of the extra data transmitted and the complexity required in the decoder, DVB subtitles often do not work nearly as well as their teletext counterparts, arriving on the screen too late, in the wrong position or colours, or not at all. Sometimes, having arrived, they are reluctant to be dismissed.
Often, both DVB and teletext subtitles are included on digital transmissions, and the viewer may select which ones to watch, but in the UK only DVB subtitles have ever been used on terrestrial, while on satellite teletext '888' subtitles have been the norm, with DVB ones being only recently introduced for the benefit of the new 'Freesat' satellite decoders which operate independently of the proprietory Sky Electronic Programme Guide and text system.
Sky digital satellite receivers ignore the DVD subtitles, and instead decode the teletext ones, displaying them in a 'fancy' proportional font rather than the old teletext dot-matrix style fixed-width font.
Here are a few split-screen photos taken of UK, French and German subtitled programmes. They were taken with a digital snapshot camera from RGB colour separations displayed on a monochrome telly and then the three exposures were combined in a graphics package.
A subtitle on Channel Four rendered by a free-to-air digital satellite receiver set for DVB (left) and as a teletext subtitle by the Sky Digibox (right)
A subtitle on France 3 rendered by a free-to-air digital satellite receiver set for DVB (left) and teletext p888 (right). The graphic used for section separators on this page is a photograph of the teletext data lines from the vertical blanking interval of a France 2 analogue satellite transmission on Atlantic Bird 3 at 5.0°W.
A subtitle on Zweites Deutsches Fernsehen rendered by a free-to-air digital satellite receiver set for DVB (left) and teletext p777 (right). The subtitle is best left untranslated.
Teletext Website Bookmarks
Hardy's UK Teletext Site
Back To Top
Alan's Home Page
Compiled by Alan Pemberton
Sheffield, South Yorkshire, England