exxos's Atari, Amiga & retro forum

As people should be aware I have been looking towards something like a Acorn RAM board.. https://www.exxosforum.co.uk/forum/viewt ... =13&t=2356

I was looking towards the sockets for the big RAM chips, in part soldering the sockets to the motherboard , so the chips can be changed assuming half of the chips are going to be faulty from China. Though there does not seem to be any reliable source for these sockets so I think that idea is a dead-end.

So we are back to the simm socket really. But as people know these are expensive. The plan was to do away with the SIMM socket and just use regular pin headers so I can test a new mini RAM board to make sure they work. Soldering RAM to the motherboard is just not a good idea as I have said previously.

I can make some small RAM boards, but I think the price to the buyer if anything is going to be higher than people buying a SIMM socket and a 4MB simm anyway.But really this is just basically reinventing the Falcon RAM board which seems little pointless. So I think the SIMM socket will just likely stay as is. While the sockets are expensive, overall it is still going to be the cheapest ram solution anyway.

I guess if people really wanted to avoid the socket, someone could maybe make a small PCB adapter to solder into the SIMM socket holes, and physically solder the same itself to the PCB. But with this being basically a bit hacky with the time involved I just don't think it is worthwhile.

The H4 rev F which is the current board design will still have the Falcon RAM board. Removing it is actually a lot of work. So on the assumption that rev F will be the final board, that will basically always stay the same for the H4 series.

Because I do not know yet what the next board revision would be, if there was a next board, the Falcon RAM board would go along with the SMT GLUE & MMU. But as this would entail a huge amount of rerouting, I would actually redesign the board differently anyway. I would likely move the GLUE over to where the MFP is, move the PLCC CPU to the front of the PCB along with the blitter. Then with the saved room under the PSU, move the MMU & simm more under the PSU so its closer to the shifter. I think the board routing would be a hell of a lot better like that, the problem being the entire board would need rerouting, and I really do not want to spend another 4 months of the year routing a new board. Also as said previously, a new board will be going a lot more SMT based. But likely that board would be a project for next year anyway. Again it depends on if people are still interested in the project by then.

Such a board will likely jump to the H5 series. we would likely have a proper 3.3V expansion bus as mentioned previously. Aside from that I do not know what else currently feature wise that series would have.. It depends how we get on with all the various other add-ons over the current year... I really want to get 16Mhz stable on the H4, so that could be a factor on the H5.. Getting this to work stable is of course a nightmare people are thinking of holding out for this revision, it may well just simply not be possible. But I do plan to get back to this mod later in the year.

Just a quick vote for keeping the simm socket. Presuming the cpld glue, mmu, etc kicks off as planned, it would be much easier to change out the 4mb simm for a 16, 32, etc etc.

PhilC wrote: 06 Jan 2020 16:52 Just a quick vote for keeping the simm socket. Presuming the cpld glue, mmu, etc kicks off as planned, it would be much easier to change out the 4mb simm for a 16, 32, etc etc.

You mean fit a 16MB simm instead of a 4MB one ? It should take 16MB, but is only going to use 4MB because of the MMU limit.

In terms of FPGA MMU, we wouldn't use DRAM at all, its not fast enough. it would have to be SRAM, so it wouldn't have a simm at all in that case.

Ok, didn't know that the dram wasn't up to the task @exxos

PhilC wrote: 06 Jan 2020 17:14 Ok, didn't know that the dram wasn't up to the task @exxos

I think it bottles out around 50-60ns... Saying that SRAM is expensive but at least currently like 10- 45ns is possible. But these are problems for another day, as it basically boils down to speed versus cost with ram.

Has anyone asked if its possible to put jumpers on the drive select controler Yamaha 2149 i think so that the internal and external drive can be swapped with a switch

Tom

Tomswork wrote: 19 May 2020 07:51 Has anyone asked if its possible to put jumpers on the drive select controler Yamaha 2149 i think so that the internal and external drive can be swapped with a switch

Unless I've misunderstood something, it already has that. I think you could just put a DPDT switch on the JP17 block to do what you're asking for.

I havent read the whole thread so forgive me if i repeat something.

Surface mount everything that moves and get the board house to pre-assemble.

For me I'd like to see 40 pin dip roms because they're so much easier to remove, wipe and reflash. We also have @cmorley's usb reflashable 27C400 emulator that can be switched by software. Keep the PLCC rom socket but have DIP as an option.

I'd also put a CPLD that does the address decode on the board.

Thank you I'm on it with a switch that make life easy

Tom

@exxos @Icky

How about this for a suggestion

Ditch the 27c4096 rom socket and stick another 68k socket next to the flashy clock socket and then give people the option to use the FC socket for either FC, rom and decoder board, etc etc?