The Sinclair QL Forum

So were there any expansion cards that actually used the RGB video line connections on the expansion connector?

Mark

Nope.

If one comes along, just move the jumpers on the board to reconnect the lines and all will be fine. I might incorporate a buffer IC to isolate the feed into the board from the feed going to the new circuitry.

So Sinclair's routing the RGB video signals to the expansion connector was a complete waste of time and resources. I can kinda understand why an expansion may need the sync signals, but I can't think of any practical and useful reason for making the RGB video signals available on the expansion connector.

I have a number of QL's. Like Derek, I don't use microdrives on my usual QL. Using a RGB lead to my Sharp Aquos LCD gives a picture that is perfectly fine. I have already isolated the +5V supply and the video feed to the modulator.
So next time I have this machine open, I may well cut the RGB tracks that go towards the expansion connector if I can get to the junction (or close by) where they split off. I may also isolate the MC1377 and associated circuitry, as I don't use the colour composite baseband video signal either.

Mark

All good steps.

The Gold Card and Super Gold Card use CSYNCL or VSYNC, I don't recall which, as a timer. It is required for memory refresh to work correctly. If the SGC had SRAM, it could be omitted. However, there would be no discernible benefit.

On a separate note, Nasta worked out the true function of that little patch board on US QLs after I sent him the layout and how it was connected.

The 15MHz clock doesn't always have a 50/50 duty cycle. The 8301 derives a 10MHz clock from the 15MHz by multiplying by two then dividing by three. The crystal gives a sine wave with a quite low level, low around 0.8V and high around 3.1V, but the 8301 expects a TTL input. The point at which a 1 becomes a 0 and vice versa can result in the clock appearing to the 8301 to have as bad as a 65/35 duty cycle. If the duty cycle of the 15 MHz clock varies too much, this results in each alternate column of pixels being a different width. This is more apparent on US QLs due to their NTSC output and larger fonts. So they used a 74HCU04 to feedback into the signal to buffer/amplify it a little. The increased swing brings the duty cycle at TTL levels into the 50/50 area.

Right now I am costing the options of either including this circuit on the video board, or of including an accurate TTL 15MHz clock. Although the effect is barely perceptible, it is a modest improvement. It was obvious enough on NTSC boards for Samsung to include the mod on every one.

I still do not know how the 8301 knows to create an NTSC or PAL signal.

[Edited to fix math typo (15*3)/2 <> (15*2)/3- Dave]

1024MAK wrote:So Sinclair's routing the RGB video signals to the expansion connector was a complete waste of time and resources.

In 1984, genlock was just becoming a plaything. Unfortunately, they didn't include a mechanism for synching the QL to an external signal - but it appears they did think at the time that some kind of video system would be produced.

My first prototype was on the expansion port, because all the signals were there. The internal version, with signal buffering and isolation, is noticeably better quality.

It was a bad decision in hindsight, I think. That said, it did provide the refresh timer for (S)GC, so that was a novel use.

Dave wrote:The 8301 derives a 10MHz clock from the 15MHz by multiplying by three and dividing by two

Is that the right way round? 15*3/2 = 22.5, but 10*3/2 = 15.

Just wondering.


Cheers,
Norm.

(15 * 2) / 3.

Fixed in original post. My typo!

Aha! You confused me by saying multiply by 3 and divide by 2.

Thanks.


Cheers,
Norm.