Over the last couple of days I put together the memory to screen interface and hooked up a 32K static memory chip (easier and cheaper to get hold of than 16K). Obviously in the absence of a anything to initialise the contents of this memory to a sensible state all I could expect to see on the screen was a random pattern. This won't be to bad though, as hopefully I'd see the familiar blocky attribute pattern over a random bitfield.
Here is a picture of the prototype as is (click here for a large version).
It was a bit of an adrenalin moment plugging it into the TV and switching on, and this is what I saw:
It was very exciting. The picture is clear, but without a proper picture it's difficult to say how sharp it is going to be. To do this I took a 32K EPROM, downloaded four loading screens from World Of Spectrum and wrote them to the start of each 8K page within the EPROM. I then substituted the EPROM in place of the RAM and with a bit of manual switching of the A13 and A14 address lines was able to switch between the four screen banks.
This gave the following four pictures:
The screen is indeed sharp and clear, with no ghosting. There is a tiny bit of 'noise', most noticeable at the boundary between dark and light colour blocks. I suspect this is either due to a bit of instability in the clock or just transient noise in the supply at the output. Just two guesses - sometime I'll try out some filtering via an inductor in the RGB output circuit, and also try out the most stable 7MHz clock I can arrange. It could just be interference picked up by my scart lead...