Little spy told me there might be a new Zeddy on the horizon

[Andy Rea]

I'm planning an uber-zeddy wih all the bells and whistles built-in.

C

Would one be requiring an Uber ULA to go with it ?



Andy

[sirmorris]

I's got one

[sirmorris]

So the first steps have been taken. I've found a bare, dead motherboard and replaced both the CPU and ULA with known workers. Success! A lovely dim K. The board is in fair shape - it's an issue 1 so not the prettiest - so I engaged the iron.

It would have been pointless doing anything more without a keyboard. I have an old external board and connector cable so I soldered forward-facing header pins to the underside connections of the ribbon headers.

Next up I...

• swapped the regulator for a '7805 compatible switching regulator' so I can remove the heatsink.
• replaced the CPU with a slightly higher-specification one simply because I could.
• added a why-wait? mod.
• fitted my Andy-ULA.

The why-wait mod was the most involved fix even then it only took 10 minutes. So firing up clkfreq.p I waited for my free 10%! Imagine my surprise when it looked like my machine was actually going _slower_ It turned out that I'd forgotten to lift R1 With the fault corrected I re-ran the test and lo! There was my expected 109%

Next up - the ULA wiring. Another 10 minute job following the clear instructions in Andy's manual. No messing around this time though - all present and correct. Happy pokes give 259% increases. Monster maze was never so hard..!

I'm now considering my next move. I suspect it will be piggy-backing RAM over the ROM. This should give me free UDG. I need to research how the various UDG schemes out there work though. It would be silly to implement something that is incompatible with existing software.

Also along the line is an internal ZXpand. This is why I wanted the heatsink removed. There will be a little problem however. The zxpand code occupies the character set data addresses, and because characters are always fetched from the internal ROM that works out nicely. But if I move the szxpand to the internal ROM ... well I'm sure you see my problem. My current thinking is that I have a number of options.

• have a 16K ROM
• keep the overlay ROM idea
• develop a hybrid ROM which lives at 0-8k and after 48k which I can do because Andy's ULA has m1-not

16K ROM at 8K would be ideal because it's simple and I could have extra room for features but that then interferes with RAM at 8K which people are starting to rely upon. The split ROM idea might be the way to go but that involves extra selection logic. Or I suppose the overlay thing may be simpler.

I hope there are more options to add to the list once I start hearing back from all you lovely people out there

[1024MAK]

There will be a little problem however. The zxpand code occupies the character set data addresses, and because characters are always fetched from the internal ROM that works out nicely. But if I move the szxpand to the internal ROM ... well I'm sure you see my problem. My current thinking is that I have a number of options.

• have a 16K ROM
• keep the overlay ROM idea
• develop a hybrid ROM which lives at 0-8k and after 48k which I can do because Andy's ULA has m1-not

16K ROM at 8K would be ideal because it's simple and I could have extra room for features but that then interferes with RAM at 8K which people are starting to rely upon. The split ROM idea might be the way to go but that involves extra selection logic. Or I suppose the overlay thing may be simpler.

Could a special "bank" switching arrangement work, where by a 16k EPROM is located in the 0 to 8k address space, with the highest EPROM address bit being the bank switching bit. Both banks contain almost the same slightly modified Zeddy ROM (so that the ROM can switch itself between banks without crashing), one of which contains the character set data and the other contains your ZXpand code?
This assumes that such an arrangement would work, and that not too many m/c programs are broken by the modified section of ROM code...
Mark

[Andy Rea]

Charlie, you can;t have executable code after 48K, m1not allows 32-48k to be used for executable code.

Andy

[siggi]

Hi Charlie,
if you would choose a memory map similar to the ZX96 you would have the chance to run programs, which were written for the ZX96 and use UDG or HIRES in upper ram (e. g. a pcb layout program or to control a model railway )

http://www.fischerkai.de/zxteam/rammap_e.htm

Bank switching at 8-16K is done by POKE 8, .., at 48-64K by POKE 9, ..

Siggi

[sirmorris]

Hmm, zx96 compatibility... That's a good thought

@Mark - Funny you should say that. I was hauling a similar thought out of my brain soup earlier. As far as I can remember the character data is selected with /refsh low. I could use that to select my ROM bank. so, a 16K EPROM with regular ROM in one half and ZXpand ROM in t'other. /refsh wired to A13. Would that do it in the simplest case?

If only there was a proper emulator which had logic level simulation of memory selects ... oh! hang on! ClockCycle! Oh.. hang on... Mac only (

C

[yerzmyey]

Of course I don't understand a word from this, but as far as I remember, there was also ZX2000, so if You want to put _everything_ in one machine....

[sirmorris]

I'd like a ZX2000 too

My aim is to have a zeddy which looks from the outside like an ordinary machine. On the inside it will be stuffed with goodness though

[Paul]

I'd like a ZX2000 too

Did you ask Kai Fischer? There might be some boards left. And it fits internally and has loads of goodies as well