Acorn Teletext adapter on B+

Arc/RPCs, peripherals, RISCOS operating system & ARM kit eg GP2x, BeagleBoard
cmjones01
Posts: 135
Joined: Fri Sep 06, 2013 2:12 pm
Location: Warsaw, Poland and Cambridge, UK

Acorn Teletext adapter on B+

Postby cmjones01 » Tue Nov 21, 2017 12:24 pm

Does anyone here have experience using the Acorn Teletext Adapter on a BBC B+? I am hoping to set up a teletext display in my office using vbit2 and teletext on a Raspberry Pi. I've got everything set up and it almost works but not quite.

IMG_20171121_131753.jpg


There's a good teletext output from the Raspberry Pi and I've built a little video buffer to feed the RPi's output into the teletext adapter after the UHF tuner. Everything seems to be working correctly inside the Acorn adapter (I've done a lot of reverse engineering in to how it works, which I'll write up at some point) but I can't get a decent display from the ATS ROM in the B+. The header line looks OK, and I see the bottom half of a page, but the top half only appears sporadically. In tuning mode it shows full signal but the time is fixed at 21:50/36, which is odd.

IMG_20171121_130618.jpg


The odd thing is that moving the ATS ROM around in the B+'s ROM sockets makes a difference to the displayed page, and sometimes pressing keys will change it too. I tried removing all other ROMs apart from the OS and BASIC but that didn't fix it.

Is the Acorn Teletext Adapter known to work with the B+? Should I fetch a model B or Master from home and try that?

Thanks!
Chris

guesser
Posts: 147
Joined: Mon Jun 26, 2006 9:21 pm

Re: Acorn Teletext adapter on B+

Postby guesser » Tue Nov 21, 2017 1:03 pm

I assume you mean the Beeb getting stuck after receiving the first header? Not that it's always that time for every test.

cmjones01
Posts: 135
Joined: Fri Sep 06, 2013 2:12 pm
Location: Warsaw, Poland and Cambridge, UK

Re: Acorn Teletext adapter on B+

Postby cmjones01 » Tue Nov 21, 2017 1:04 pm

I know it's bad manners to reply to one's own posts, bu I had a brainwave. Though the Acorn Teletext adapter and ATS seem to be designed to handle 16 lines of teletext information, I wondered if that in practice they couldn't. I edited the source to the Raspberry Pi's teletext program (reducing HEIGHT in teletext.c) and lo and behold it works much better. Still not quite right, but at least I have a full screen of teletext now. Curiouser and curiouser.

Chris

cmjones01
Posts: 135
Joined: Fri Sep 06, 2013 2:12 pm
Location: Warsaw, Poland and Cambridge, UK

Re: Acorn Teletext adapter on B+

Postby cmjones01 » Tue Nov 21, 2017 1:05 pm

guesser wrote:I assume you mean the Beeb getting stuck after receiving the first header? Not that it's always that time for every test.

It's more strange than that - the Beeb isn't stuck, the time at top right updates continuously, and I get occasional flickers of the rest of the page.

guesser
Posts: 147
Joined: Mon Jun 26, 2006 9:21 pm

Re: Acorn Teletext adapter on B+

Postby guesser » Tue Nov 21, 2017 1:05 pm

aha! I should have thought of that, I have exactly the same issue with the Volex adapter for the ZX Spectrum that I'm currently reverse engineering :lol:

guesser
Posts: 147
Joined: Mon Jun 26, 2006 9:21 pm

Re: Acorn Teletext adapter on B+

Postby guesser » Tue Nov 21, 2017 1:13 pm

Bear in mind that editing teletext.c to generate fewer lines will have side effects (though probably irrelevant for the beeb) because vbit2 hard codes the number of lines for the purposes of counting frames.

This is something that should probably be settable via the config file in the short term, and in the long term it needs a different interface to raspi-teletext to allow it to use only particular lines... but I digress! :D

cmjones01
Posts: 135
Joined: Fri Sep 06, 2013 2:12 pm
Location: Warsaw, Poland and Cambridge, UK

Re: Acorn Teletext adapter on B+

Postby cmjones01 » Tue Nov 21, 2017 1:23 pm

guesser wrote:aha! I should have thought of that, I have exactly the same issue with the Volex adapter for the ZX Spectrum that I'm currently reverse engineering :lol:

Ooh, interesting. I wonder how similar it is to the Acorn one? The Acorn one basically loads each line of teletext data raw into an internal 1K RAM, then generates an interrupt to the Beeb. The Beeb than has less than 20ms to read the data out of RAM, clear it (I think), and process it. If it doesn't manage it, there's a data overrun flag generated by the adapter which tells the Beeb it's missed something.

Chris

guesser
Posts: 147
Joined: Mon Jun 26, 2006 9:21 pm

Re: Acorn Teletext adapter on B+

Postby guesser » Tue Nov 21, 2017 1:31 pm

Completely different really, the Acorn one is as you say like a TV teletext decoder and the beeb just grabs the page data from it.
EDIT: unless I've misunderstood how the acorn one works...

While I've not probed out all the logic side of things yet so can't say when and how often signals occur, the speccy one stores incoming teletext packets into a 1kB SRAM and triggers an NMI. The speccy then loops over the buffer and copies the packets into its own RAM and decodes them.
The logic side of the interface is very sparse, just a ROM, RAM, and custom gate array. Nothing teletext specific at all except for a clock regen and data slicer IC on the tuner board.

cmjones01
Posts: 135
Joined: Fri Sep 06, 2013 2:12 pm
Location: Warsaw, Poland and Cambridge, UK

Re: Acorn Teletext adapter on B+

Postby cmjones01 » Tue Nov 21, 2017 1:36 pm

guesser wrote:Completely different really, the Acorn one is as you say like a TV teletext decoder and the beeb just grabs the page data from it.
While I've not probed out all the logic side of things yet so can't say when and how often signals occur, the speccy one stores incoming teletext packets into a 1kB SRAM and triggers an NMI. The speccy then loops over the buffer and copies the packets into its own RAM and decodes them.
The logic side of the interface is very sparse, just a ROM, RAM, and custom gate array. Nothing teletext specific at all except for a clock regen and data slicer IC on the tuner board.

That sounds extremely similar - the data in the Acorn RAM is raw teletext packets which the Beeb then has to deal with. The Speccy software would have a similar rush to decode things in order not to miss the next frame's packets, so it's interesting that you're seeing the same problems as I am.

The later Acorn/GIS decoder uses a later chipset which does the decoding in hardware, like the Morley one. Actually, I've got a Morley adapter I should try out too.

Chris

guesser
Posts: 147
Joined: Mon Jun 26, 2006 9:21 pm

Re: Acorn Teletext adapter on B+

Postby guesser » Tue Nov 21, 2017 1:43 pm

Yeah I've gone back and read up on it again. Faulty memory, I thought it decoded the headers on-board. :oops:

It is indeed rather similar theory of operation, just much simpler interfacing to the speccy than tube!

It has the same TDA2541 and things on the tuner board. Is the SAW filter a RW153A? Will be interesting to compare RF circuits, I expect both to be very similar to the datasheet reference design :)

guesser
Posts: 147
Joined: Mon Jun 26, 2006 9:21 pm

Re: Acorn Teletext adapter on B+

Postby guesser » Tue Nov 21, 2017 1:45 pm

cmjones01 wrote:The later Acorn/GIS decoder uses a later chipset which does the decoding in hardware, like the Morley one. Actually, I've got a Morley adapter I should try out too.

Yeah, that's the one I was thinking of. I believe it interfaces with the teletext IC using I²C?

cmjones01
Posts: 135
Joined: Fri Sep 06, 2013 2:12 pm
Location: Warsaw, Poland and Cambridge, UK

Re: Acorn Teletext adapter on B+

Postby cmjones01 » Tue Nov 21, 2017 1:51 pm

guesser wrote:It has the same TDA2541 and things on the tuner board. Is the SAW filter a RW153A? Will be interesting to compare RF circuits, I expect both to be very similar to the datasheet reference design :)

The SAW filter is an SW153A in my Acorn adapter, driven by an SL1430. There's a handy link, LK3, which allows you to disconnect the video from the TDA2541, so I've used that to inject composite video from the RPi via a little amplifier to bring it up to the 2V pk-pk that the SAA5030 wants.

cmjones01
Posts: 135
Joined: Fri Sep 06, 2013 2:12 pm
Location: Warsaw, Poland and Cambridge, UK

Re: Acorn Teletext adapter on B+

Postby cmjones01 » Tue Nov 21, 2017 1:51 pm

guesser wrote:
cmjones01 wrote:The later Acorn/GIS decoder uses a later chipset which does the decoding in hardware, like the Morley one. Actually, I've got a Morley adapter I should try out too.

Yeah, that's the one I was thinking of. I believe it interfaces with the teletext IC using I²C?

That's the job, yes.

guesser
Posts: 147
Joined: Mon Jun 26, 2006 9:21 pm

Re: Acorn Teletext adapter on B+

Postby guesser » Tue Nov 21, 2017 2:08 pm

cmjones01 wrote:The SAW filter is an SW153A in my Acorn adapter, driven by an SL1430. There's a handy link, LK3, which allows you to disconnect the video from the TDA2541, so I've used that to inject composite video from the RPi via a little amplifier to bring it up to the 2V pk-pk that the SAA5030 wants.


I think SW153A is the same part essentially. The SL1430 is interesting, that is what the example schematic in my Newnes Television and Video pocket book shows. In the Volex adapter that amplifier is omitted. Presumably the output from the tuner used (mullard u343) doesn't require amplification while others do.
What tuner does the Acorn have? I found mine in a databook on archive.org https://archive.org/details/MullardBook ... Tuners1987

cmjones01
Posts: 135
Joined: Fri Sep 06, 2013 2:12 pm
Location: Warsaw, Poland and Cambridge, UK

Re: Acorn Teletext adapter on B+

Postby cmjones01 » Tue Nov 21, 2017 2:15 pm

guesser wrote:What tuner does the Acorn have? I found mine in a databook on archive.org https://archive.org/details/MullardBook ... Tuners1987

It's a Mullard U341 (Philips 12NC part number looks like 2122 127 41491 but the label's a bit scuffed).

cmjones01
Posts: 135
Joined: Fri Sep 06, 2013 2:12 pm
Location: Warsaw, Poland and Cambridge, UK

Re: Acorn Teletext adapter on B+

Postby cmjones01 » Fri Nov 24, 2017 10:24 pm

A quick update on this one: I got my Morley teletext adapter running and that works fine with the Raspberry Pi. It uses a newer chipset that does all the page capture and decoding in hardware so the Beeb's job is easier.

Having thought about it a bit more, I have a theory as to why the Acorn adapter doesn't work well with the Raspberry Pi. According to the broadcast teletext specification, the only time the number of a page is sent is in row 0, the page header. The page number is then implicit in all rows until the next page header. That means that if the decoder misses a page header, perhaps because it has too much data to deal with or isn't aware of 16-line teletext broadcasts, it can end up taking rows from different, unrelated pages and cobbling them together on screen until it actually catches another page header. That matches the behaviour I was seeing. Time for a new version of ATS which addresses this problem? ;-)

Chris


Return to “hardware”

Who is online

Users browsing this forum: No registered users and 6 guests