exxos's Atari, Amiga & retro forum

Well, I'm getting somewhere with the RAM expansion.

I made my own footprint for the DIN socket, which will be to provide a ZX Spectrum+ 128K compatible RGBI output on the back of the machine.

Here's the first render from KiCAD. I've got both pads for directly soldering the surface mount SRAM on the board and a socket to put the daughter board into. These boards have already been built and one is going to have the component soldered on-board by PCBWay so I don't have to.

The paper mock-up looks to fit correctly at least.

Bother! Just realised that I have a logic error in my bus control logic. The RAM won't work and I risk halting the whole machine in another state.

It's a good thing I'm checking, double checking and triple checking.

I need to check to see if the GALs pull the lines high or let them drift. I know that I can tri-state outputs, but that needs a control line and you can't use OR operations in the equation. Could be tricky.

OK, it looks like my whole approach is flawed, so I'm going to have to backtrack and not have the SRAM chip accessible from the expansion card.

This simplifies things a great deal. I can still have the address latching on this board and pass the valid address to the daughter board but all the logic and memory for any expansions will have to be on there. This will cut the chip count by three (a GAL, an LS04 and an LS125) and I'll only ever use a quarter of the memory. C'est la vie.

Well, I think that I've got the simplified version of the board completed.

Daughter boards can use the address latching and transceivers on this board but otherwise everything else is being passed through to the expansion connector. Any expansions will have to provide their own memory, address decode logic etc. but at least some functionality won't need to be duplicated.


After checking a cut-out of the board in the Apricot F2 I discovered that I needed to change the dimensions of the board slightly and add a couple of cut-outs. I've also moved the surface mount SRAM pads to the back of the board if for nothing else to get rid of the warnings when I do a check.

I had to put the 74LS04 back on the board as I'd forgotten about a couple of lines it was inverting for me. I used the opportunity to make sure that the /SELECTED line was not being actively pulled high, as a GAL will want to do, and hence push it through the 74LS04 as well. The /SELECTED line needs to be able to be pulled low by the board on the expansion connector.

During all this checking and re-checking I also discovered that I'd forgotten to replace a couple of the signals on the expansion connector, so things would not have worked well (OK, at all!).

Anyway, I think that I've now debugged the board and I'm running Freerouting. I've decided that as the main motherboard seems to be a two layer board I'll try this as a two layer board, as it will be FAR cheaper. It's a bit of a risk but the machine isn't running quickly. The CPU is running at 4MHz but the memory bus is really running at a quarter of this given the multiplexing and the enforced one wait state on every memory or I/O access.

Well, the boards are due to be delivered tomorrow so I've been checking my stocks and I have enough of everything to build more than one up.

The pre-made SRAM daughter board is still waiting to ship from PCBWay and has been since last Wednesday. I don't know what the hold up is as they finished soldering the SRAM chip and capacitor on that board on Tuesday. Still, it just means that I'll have to try soldering a surface mount SRAM chip on the main board directly.

Sounds like you're getting on great with this!

Will it be visiting the Centre for Computing History?

I'd love to come and see it but the 9th/10th October don't work for me

rubber_jonnie wrote: ↑Sun Sep 12, 2021 4:56 pm Sounds like you're getting on great with this!

Will it be visiting the Centre for Computing History?

I'd love to come and see it but the 9th/10th October don't work for me

Yes, I'm planning this on my Saturday set up. We'll see if the board actually works.

stephen_usher wrote: ↑Sun Sep 12, 2021 5:31 pm

rubber_jonnie wrote: ↑Sun Sep 12, 2021 4:56 pm Sounds like you're getting on great with this!

Will it be visiting the Centre for Computing History?

I'd love to come and see it but the 9th/10th October don't work for me

Yes, I'm planning this on my Saturday set up. We'll see if the board actually works.

Hmm, talking to my other half, might be able to make the 9th work