Teletext is 40 years old

[Sysman]

A salute to the forerunner of the internet world wide web

For those of us that had a teletext set, it was the zenith of cool. In a day when the internet world wide web was still a twinkle in the eyes of a CERN engineer, it really was the world at your fingertips. Hundreds of pages of news and information at the touch of a button, controlled by a computer in London with the RAM of a ZX81.

[BrilloPad]

For those of us that had a teletext set, it was the zeity of cool. In a day when the internet world wide web was still a twinkle in the eyes of a CERN engineer, it really was the world at your fingertips. Hundreds of pages of news and information at the touch of a button, controlled by a computer in London with the RAM of a ZX81.

FTFY

[NickFitz]

Another excuse for me to remind people that I once had a job writing control software for Teletext transmission systems

[Sysman]

Another excuse for me to remind people that I once had a job writing control software for Teletext transmission systems

The article says "controlled by a computer in London with the RAM of a ZX81" but does that just mean the same amount as a ZX81 had?

Do you know what that computer and OS were?

[NickFitz]

The article says "controlled by a computer in London with the RAM of a ZX81" but does that just mean the same amount as a ZX81 had?

Do you know what that computer and OS were?

The systems made at that company were 6809 based, and though I can't offhand remember how much RAM a deployed system would have, I'm pretty sure our development systems had 64K. By the mid-80s they were also starting to make 8086 based boards, but I don't know how much they would have had.

The hardware that did the work of putting the Teletext data into the TV signal was called the Channel Insertion Unit (CIU). It was basically a thing that detected the vertical sync of the outgoing transmission and generated an interrupt. On that interrupt, our code would grab the lines that were due to go out on that frame from the data store, called the magazine, and write them into the CIU's FIFO buffer; it then took care of streaming them into the appropriate lines of the outgoing TV signal.

For development purposes, we had a signal generator that just sent blank frames. This meant that you could tell it to insert the rows in the middle of the frame, rather than at the top. If you also just sent loads of copies of the same row, you could then see it as a pattern of vertical bars in the middle of the frame, which was the binary representation of the data; this allowed you to check that you were formatting stuff correctly.

The CIU was also capable of chucking out rows for the entire height of the frame, known as full-frame teletext. This was used to get much higher bandwidth in situations where you weren't actually sending any TV content, which was the case for the displays that used to be found in railway stations, airports, and suchlike places around the world.

The control software was all written in FORTH, and assembly language for time-critical stuff like the vertical blanking interrupt handler that primed the CIU each frame. The magazine was stored on floppy disks (some of the systems were still using 8" floppies, though 5 1/4" ones were rapidly replacing them) as standard FORTH blocks. (FORTH is its own operating system and just treats the entire disk as sequential 1KB blocks accessed by numerical index, rather than bothering with irrelevancies like files and directories.)

So although the systems had a good bit more RAM than the average ZX81, and of course the CIU hardware, they weren't that far off. After all, two chaps left Sinclair and made a machine almost the same as the ZX81, but with a 6809 processor and FORTH: the Jupiter Ace. I suspect the build quality of our kit was a good bit higher, as it was intended to operate 24/7/365 for many years, but other than that the Ace and our kit had a lot in common

[NickFitz]

Oh, the company also made Teletext editing terminals, with extra features such as a 2x3 matrix of keys for toggling the individual blocks of the "graphics" characters, and function keys to insert control codes for features like double-height and colour codes. These terminals were connected to the transmission unit's RS232 serial port.

They had some capability for storing several pages locally in the short-term, but only in RAM. Saving was done on the transmission unit, and once a page was marked as active it immediately joined the queue to be transmitted when its turn came up. So the people who operated these terminals had to avoid doing things like drawing blocky pictures of cocks or typing "BUM" in double-height magenta letters, then accidentally adding it to the magazine

[Churchill]

Sorry, old chap, the Ace uses a Z80.

Z80A - sorry old chap :-)

[NickFitz]

Sorry, old chap, the Ace uses a Z80.

DOH!

It does indeed - dunno where I got the idea it used a 6809 from, other than the fact that the 6809 was ideal for FORTH, having two hardware stacks. I think I was mixing it up with the Tandy Color Computer and the Dragon 32, which did have a 6809. I stand corrected

Anyway, they're still 8-bit processors running FORTH