ZX81 on modern TV

[Gulliver]

Hi, I've searched and not found an answer. I have two Sinclair ZX81's, one is original and the other has been modified to give a composite video output (mod found on a ZX81 message board).

Lots of people seem to have problems making their Zeddies work with flat screen TV's so I was wondering if the good people on this forum have any ideas? Neither of mine work with my LCD television but I have somewhat better results with older CRT televisions.

[cmjones01]

This is always going to be difficult. In my memory of looking at the ZX81's video output on a scope, it has no 'back porch' at black level in between the line sync pulse and the (white) screen content. It just goes straight from sync to white, because it's a cheap-and-cheerful design. That can mean that modern TVs, in their standards-compliant way, carefully clamping their black level to the level just after the sync pulse, see the whole picture as black.

If I had to fix it, I'd put together a sync separator with a monostable triggered from it which forced the video output to a nominal black level for a couple of microseconds after the rising edge of each sync pulse. The trusty LM1881 would be a good start.

Caveat: this is all based on old memories of fiddling with ZX81 video output. I do remember that the picture tended to be very dark on monitors, which is exactly what I'd expect from this problem.

[richrussell]

I seem to recall I got mine working by feeding the ZX81's modulator output into an VCR with a SCART connection to the television. The early 1990s video is much more tolerant of the ZX81's signal, and happily converts it to a composite signal acceptable by my modern LG LCD television. This was a few years back though.

The picture is still terrible because of it's low resolution and poor stability, but it's good enough for a game of Mazogs Or would be if I didn't discover my ZX81's keyboard ribbon had shattered.

[neon indicator]

Cheap LM1881 IC and a few resistors and caps and maybe CMOS switch. I've used them to clean 1.2GHz ATV receiver sync as those are often terrible and picture rolls.

Maplin even sell them.

[Gulliver]

You can still buy brand new ZX81 keyboards as the ribbon connectors are known to fail. For that matter you can buy new Z80A processors at Maplin too There's little excuse for not trying to use a ZX81 except the problem with modern TV's.

Yes it is the 'back porch' problem. At home I have an AOC LCD television which gives a reasonable contrast but a very unsteady picture. I've tried routing the ZX81 via a DVD recorder and a 2001 vintage VCR but to no avail. I don't have the space for more equipment, the bringing in of an old VCR (I have several in storage) would be grounds for divorce!

One of my ZX81's is modified to give composite output and it works for 30 minutes, but as the ULA gets hot the picture breaks up.

Can you be more specific about what I need to do with the LM1881 ?

[cmjones01]

The LM1881 data is here, on National Semiconductor's website:

http://www.national.com/mpf/LM/LM1881.html

A very handy chip, it'll extract sync pulses from a video signal with very few external components. It has an output called 'burst/back porch' on pin 5. I'd use this to switch on an NPN transistor or N-channel MOSFET to ground, with a resistor from its collector/drain to the video line chosen empirically to drag the video signal down to about 30% its peak white voltage.

Of course, I don't know if the LM1881 actually produces a useful back porch pulse if the input doesn't have a back porch. That is left, as they say, as an exercise for the reader...

[julie_m]

It would be a simple enough matter to add a monostable to produce the correct length pulse, if necessary.

[SiriusHardware]

Excuse my ignorance, but what is the correct length for the 'back porch' segment following the sync pulse: And does it only follow the line sync, frame sync or both? (In the UK here, by the way)

If it comes after the line sync, I take it that inserting some black level right afterwards would make a narrow strip of the left hand edge of the background white area disappear into black?

-Graham (currently looking at a dim grey picture from my ZX81 on a modern Sharp flatscreen).

[Amraduk]

Hello Graham,

Originally Posted by SiriusHardware

Excuse my ignorance, but what is the correct length for the 'back porch' segment following the sync pulse:

It follows the line sync pulse and is 5.65uS in duration.

And does it only follow the line sync, frame sync or both? (In the UK here, by the way)

You will only see it on lines that contain active video, provided the video isn't at black level!"

If it comes after the line sync, I take it that inserting some black level right afterwards would make a narrow strip of the left hand edge of the background white area disappear into black?

If you mean that it would be timed to start at the end of the back porch, yes it would, provided the video wasn't at black level. However, I though the idea was to provide a back porch where none exists, i.e. in the video o/p of the ZX81. As most, if not all TVs are over-scanned horizontally (and probably vertically as well!), it will be outside the CRT's picture area anyway, and not seen!

Regards,

Dave.

[SiriusHardware]

I see what you mean - if the whole of any given line happens to be black, then the remainder of the line after the 5.65uS of back porch will be at the same level so it should be impossible to see where the back porch ends and the line begins, so to speak.

OK, next question: Would it matter if the frame sync had 5.65uS of back porch inserted after it as well? It would obviously be simpler to make a little circuit which inserted some back porch after ANY sync pulse rather than having to distinguish between frame and line sync.

Re: Overscan, no, don't forget that the ZX81's output is 4:3 and the vast majority of modern flatscreen TVs are widescreen, so the one thing you aren't going to get is horizontal overscan

I'll have to try filling the whole screen with graphic characters to see where the printable area of the screen begins, but I imagine the beginning of the printable area will be well away from the left hand edge of the white background, enough to be able to sacrifice 5.65uS from the edge anyway.

Actually, I've just thought of a major snag with all of this - video isn't the only thing the ULA pin in question is used for. It is also used to output the tape SAVE signal and read the tape LOAD signal, so although the approach above would probably fix the video problem, it would also break the operation of the tape interface, since it is possible that a simple analogue circuit would interpret the tape data stream as a stream of sync pulses.

Any solution would therefore probably have to be at least partly digital, ie, it would have to monitor specific system events in order to definitely 'know' when to insert a back porch clamp, and to remain hands-off at all other times.

[Amraduk]

Originally Posted by SiriusHardware

I see what you mean - if the whole of any given line happens to be black, then the remainder of the line after the 5.65uS of back porch will be at the same level so it should be impossible to see where the back porch ends and the line begins, so to speak.

You got it! That applies in the UK, in the US (not sure about other countries) they have what is termed 'pedestal' or 'lift', where the active part of the line has a black level about 10% above blanking level, so in that case, even with video at black level, you would see the end of the back porch!

OK, next question: Would it matter if the frame sync had 5.65uS of back porch inserted after it as well? It would obviously be simpler to make a little circuit which inserted some back porch after ANY sync pulse rather than having to distinguish between frame and line sync.

No, it wouldn't matter but it wouldn't make any difference either, because, during the frame sync period, the video is already at blanking level, ignoring any test signals or teletext data there may be during that period!

Re: Overscan, no, don't forget that the ZX81's output is 4:3 and the vast majority of modern flatscreen TVs are widescreen, so the one thing you aren't going to get is horizontal overscan

I was thinking about tube-based TVs and completely forgot about modern flat screen TVs. In any case, flat screen TVs work in a different way, the signal is processed so that only the active video portion of a line is fed to the display, there is no 'fkyback' in the conventional sense.

I'll have to try filling the whole screen with graphic characters to see where the printable area of the screen begins, but I imagine the beginning of the printable area will be well away from the left hand edge of the white background, enough to be able to sacrifice 5.65uS from the edge anyway.

I think the 'printable' area of the screen will probably extend to the edges of the viewable area, though I'n not certain about that. Of course, with a 4:3 aspect ratio signal, how close to the edges of the screen it would go, would depend on how the TV has been set up. It would also depend on how close the the start of a line (and how close to the start of a field) that the ZX81 places any data.

Actually, I've just thought of a major snag with all of this - video isn't the only thing the ULA pin in question is used for. It is also used to output the tape SAVE signal and read the tape LOAD signal, so although the approach above would probably fix the video problem, it would also break the operation of the tape interface, since it is possible that a simple analogue circuit would interpret the tape data stream as a stream of sync pulses.

I'm not familiar with the ZX81 but with the scenario as you describe, it probably would affect the tape interface.

Any solution would therefore probably have to be at least partly digital, ie, it would have to monitor specific system events in order to definitely 'know' when to insert a back porch clamp, and to remain hands-off at all other times.

Is it not possible to get at the video signal itself, after any digital processing by the ZX81, and add the back porch there, although that might blank data that should be visible?

Regards,

Dave.

[SiriusHardware]

Originally Posted by Amraduk

I was thinking about tube-based TVs and completely forgot about modern flat screen TVs. In any case, flat screen TVs work in a different way, the signal is processed so that only the active video portion of a line is fed to the display, there is no 'fkyback' in the conventional sense..

Well, actually, the ZX81 works very well with old (and I suppose I mean OLD) B&W CRT TVs which give a perfectly bright, clear picture and, with AFC, cope with the modulator's tendency to drift a little over time. It's this modern stuff which is not prepared to tolerate them.

Originally Posted by Amraduk

I think the 'printable' area of the screen will probably extend to the edges of the viewable area, though I'm not certain about that.

That certainly was true of the earlier MK14 and its add-on video interface, (both of which I also own). There, the whole viewable area displayed significant information, ie, there was no thought or allowance given to the fact that most televisions had a significant amount of overscan, so to use it you had to butcher the height/width settings on the TV to make all of the edges of the raster visible - otherwise you couldn't see/read the stuff at the edge.

Sinclair presumably noticed this and, on the ZX81 and especially the Spectrum, made such exceptionally generous allowance for overscan that the printable area ended up being a relatively small area in the middle of the raster surrounded by a huge non-printable border. So chopping a bit off the left hand side of the background area would be an acceptable loss.

Originally Posted by Amraduk

Is it not possible to get at the video signal itself, after any digital processing by the ZX81, and add the back porch there, although that might blank data that should be visible?

Like most of Clive's creations, the ZX81 is a minimalist masterpiece. The video signal emerges from the ULA pin, travels along a short wire and into the modulator. There is no intervening buffer or other convenient intercept point. Of course, this makes it pretty difficult to alter the level of any portion of the video signal because any circuit which does so does it in direct opposition to the efforts of the ULA pin to maintain what it thinks is the correct level. Put another way: Forcing the white output down to black level, even for a short time, could possibly damage the ULA.

So at a minimum, we'd need a unity-gain non-inverting buffer circuit placed in between the ULA pin and the modulator input. With that in place we could impose a back porch segment on the output from the buffer without placing any stress on the ULA's output, and, as a bonus, said imposition would no longer affect the tape load/save operation.

I should say, incidentally, that very late versions of the ZX81 were fitted with a revised ULA which addressed this issue - the revised ULA presumably generates the missing back porch signal - but finding those ULAs now (or indeed any working ZX81 ULA, early or late) is becoming very difficult.

[Amraduk]

Originally Posted by SiriusHardware

That certainly was true of the earlier MK14 and its add-on video interface, (both of which I also own).

Actually, I meant the 'printable' area of the TV, rather than the ZX81!

Originally Posted by Amraduk

Is it not possible to get at the video signal itself, after any digital processing by the ZX81, and add the back porch there, although that might blank data that should be visible?

Like most of Clive's creations, the ZX81 is a minimalist masterpiece. The video signal emerges from the ULA pin, travels along a short wire and into the modulator.

Ah, I see.

There is no intervening buffer or other convenient intercept point. Of course, this makes it pretty difficult to alter the level of any portion of the video signal...

So at a minimum, we'd need a unity-gain non-inverting buffer circuit placed in between the ULA pin and the modulator input.

Yes, that would be the way to go.

Regards,

Dave.

[richrussell]

It might be cheaper and simpler in the long run to hold on to a 1970s/80s/early 1990s TV for use with computers of that era

My ZX81 and Spectrum (as well as my Atari ST and Commodore Amiga if I use their modulator outputs) work well on my other TV - a Sony Trinitron from 2001. It is really only when the TVs started to use framebuffers and the like to process the video signal (ie for display on something that isn't a CRT) that the problems started to occur.

[SiriusHardware]

Originally Posted by richrussell

It might be cheaper and simpler in the long run to hold on to a 1970s/80s/early 1990s TV for use with computers of that era

My ZX81 and Spectrum (as well as my Atari ST and Commodore Amiga if I use their modulator outputs)

So, um, do you collect computers? (kidding!). My Amiga was the only one I sold when it was still a current model. I found the ST (which, like you, I still have) a much nicer machine to programme despite it having no hardware graphics assistance to speak of (no hardware sprites, etc).

I've just scrounged a nice little 9" 1980s Sanyo B&W portable which is perfect for the ZX81, but it's still a pain to have to have a specific TV for this one computer.

[richrussell]

Off topic - but I used to do freelance games programming in 68000 on the ST and Amiga (when I was 16!). Some of my 6502 C64 code is in commercial games published by Gremlin and Ocean.

Sadly all of my old computers are in a cupboard and don't come out to play very often. All can be emulated perfectly well on my fairly low powered laptop, so it's not worth the effort of getting them out.

[SteveSmit]

Excuse me for butting in a little here, but could that original issue apply to the original breadbin C64 too? I just ask because I bought one, but couldn't get it working on my modern TV, all I got was black. I thought the machine was faulty somehow but looking at the symptoms being talked about here I'm starting to wonder.

[Gulliver]

Originally Posted by SteveSmit

Excuse me for butting in a little here, but could that original issue apply to the original breadbin C64 too? I just ask because I bought one, but couldn't get it working on my modern TV, all I got was black. I thought the machine was faulty somehow but looking at the symptoms being talked about here I'm starting to wonder.

It is entirely possible that other computers designed for older televisions will have troubles on modern equipment. I have a C64 but it's not been fired up in a long time, ditto my Atari ST - neither since the mid 1990's. But I have read about people having problems with 64's and Spectrums.

Getting hold of and using an old TV isn't really an option for me, as I live in a flat which is already quite filled with hi-fi and computer equipment...my wife would quite possibly see a CRT as grounds for divorce! In any case I've tried my two ZX81's (one modded for composite output, one not) on various CRT's 10-15 years old and even with those the results are far from uniform.

I am beginning to think that the way to go is a UHF tuner with either composite or VGA output (or both).

[SiriusHardware]

Originally Posted by SteveSmit

Excuse me for butting in a little here, but could that original issue apply to the original breadbin C64 too? I just ask because I bought one, but couldn't get it working on my modern TV, all I got was black. I thought the machine was faulty somehow but looking at the symptoms being talked about here I'm starting to wonder.

The C64 has colour output, therefore (I would guess) it has to have a place for the colour burst signal to go - in other words, it must have a back porch present in the correct place between the sync and the beginning of the line luminance information. I suppose it may be possible that the back porch is not at the level expected by modern TVs. Can you scope the video input to the modulator? Does it look normal?

This problem we're discussing re: The ZX81 occurs because the most common version of the ZX81 ULA did not generate a back porch at all on the video signal.

[cmjones01]

Originally Posted by Gulliver

I am beginning to think that the way to go is a UHF tuner with either composite or VGA output (or both).

Such a tuner won't necessarily fix the problem. A dumb tuner which produces a composite output will reproduce more or less exactly what the ZX81 feeds into its modulator: a video signal with no back porch, so you'll be back where you started. A smarter tuner with a VGA output will involve some clever processing to change the scan rate (and decode the colour, if it's present) so this may or may not re-insert a back porch or, indeed, obediently set the black level to white in exactly the same way as a modern TV does, resulting in the same problem.

The only sure-fire way to sort this out is to insert a back porch into the video signal as it comes out of the ZX81.

I think the ZX81 is unusual amongst home computers in producing such a deficient video signal. Most other machines were either colour, and therefore had to have a back porch for their colour burst, or had a black background to the screen anyway so this problem didn't arise, or were simply more deluxe designs which did a better job of generating the video signal.